Joint analysis of genetic Joint analysis of genetic
and physical interactions and physical interactions
in in
S cerevisiaeS cerevisiae
Igor UlitskyIgor UlitskyRon Shamir labRon Shamir lab
School of Computer ScienceSchool of Computer ScienceTel Aviv UniversityTel Aviv University
Genetic interactions (GI)Genetic interactions (GI)
ldquoldquoAggravatingrdquo interactionsAggravatingrdquo interactions
The observed phenotype is The observed phenotype is worseworse than what we would ldquoexpectrdquothan what we would ldquoexpectrdquo
Synthetic lethalitySynthetic lethality joint deletion of joint deletion of two nonessential genes two nonessential genes lethal lethal phenotypephenotype
Synthetic sicknessSynthetic sickness joint deletion of joint deletion of two nonessential genes two nonessential genes slow slow growthgrowth
GI availabilityGI availability
Systematically mapped by SGA and Systematically mapped by SGA and dSLAMdSLAM
Available for asymp200 gene queries and Available for asymp200 gene queries and asymp4500 targets in asymp4500 targets in S cerevisiaeS cerevisiae
GI network 13632 interactions GI network 13632 interactions 2682 genes2682 genes
Physical interactions (PI)Physical interactions (PI)
Protein-protein interactionsProtein-protein interactions Y2HY2H TAPTAP
Protein-DNA interactionsProtein-DNA interactions ChipChip22 experiments experiments
PI network 68172 interactions 6184 PI network 68172 interactions 6184 proteinsproteins
GI Analysis SpectraGI Analysis Spectra
Genetically interacting homologs tend to Genetically interacting homologs tend to exhibit ldquocompensationrdquo on the expression exhibit ldquocompensationrdquo on the expression levellevel R Kafri R Kafri et alet al 2005 2005
Genetically interacting proteins frequently Genetically interacting proteins frequently have a similar foldhave a similar fold O Dror O Dror et alet al 2007 2007
Genetic interactions can be used to Genetic interactions can be used to delineate regulatory pathwaysdelineate regulatory pathways RP Onge RP Onge et alet al 2007 2007
Joint analysis of GI and PIJoint analysis of GI and PI
Motivation Motivation Identifying pathwaysIdentifying pathways Connect pathways with phenotypesConnect pathways with phenotypes Understand system features (robustness Understand system features (robustness
essentialityhellip)essentialityhellip)
Initial analysisInitial analysis Proteins close in the GI network slightly more Proteins close in the GI network slightly more
likely to physically interact (Tong et al 2004) likely to physically interact (Tong et al 2004) PI hubs likely to be GI hubs (Ozier et al PI hubs likely to be GI hubs (Ozier et al
2003)2003)
BPMsBPMs
Kelley and Ideker 2005 Kelley and Ideker 2005 modeling ldquoexplanationsrdquo for GIsmodeling ldquoexplanationsrdquo for GIs
Within-pathway model Between-pathway model
GI
PI
Kelley amp Ideker conclusionsKelley amp Ideker conclusions
40 of the GIs can be explained by physical 40 of the GIs can be explained by physical modelsmodels
Between-pathway models explain x 35 times Between-pathway models explain x 35 times more GIs than within-pathway modelsmore GIs than within-pathway models
Models capable of predicting protein functionsModels capable of predicting protein functions
BPM rationaleBPM rationale
Between pathway Between pathway models suggest models suggest redundancyredundancyAlternative pathsAlternative paths Independent KOs Independent KOs
little effectlittle effect Joint KOs in both Joint KOs in both
severe effectsevere effect
Viable phenotypeViable phenotype
Lethal phenotype
BPM HuntingBPM Hunting
Look for pairs of pathways with GI Look for pairs of pathways with GI evidence for bufferingevidence for buffering
A pathway ndash a A pathway ndash a connectedconnected subnetwork in subnetwork in of PIsof PIs
Model scoring based on the density of GIsModel scoring based on the density of GIs
Log-likelihood scores accounting for GI Log-likelihood scores accounting for GI degreesdegrees
1 2
1 2
( )1 2 BPM1 2
1 2 null ( )
( ) (1 )(1 ( ))( | )( ) log log
( | ) ( ) (1 )(1 ( ))a b V V
genetica b a ba b V V
I a b I a bP V V MS V V
P V V M r I a b r I a b
AlgorithmicsAlgorithmics
High-scoring seedsHigh-scoring seeds Finding heavy bicliques Finding heavy bicliques
with connectivity with connectivity constraintsconstraints
Seed optimizationSeed optimization Greedy search Greedy search
maintaining maintaining connectivityconnectivity
Significance filteringSignificance filtering
Adding ldquopivot proteinsrdquo to Adding ldquopivot proteinsrdquo to BPMsBPMs
Pathway bifurcation Redundant sub-complexes
Pivot proteinsPivot proteins
Physically connected to both pathways Physically connected to both pathways in the BPMin the BPM
Connection significant given the Connection significant given the general network degreesgeneral network degrees
Analysis flowAnalysis flow
13632 genetic interactions
68172 physical interactions
140 models
124 pivot proteins in 40 models
BPM analysis
Pivot extraction
Physiological characteristics
bullEssentialitybullPhenotypesbullProtein AbundancebullCodon adaptation indexbullmRNA half-lifebullPhoshorylation
BPMs as functional modulesBPMs as functional modules
Functional enrichment analysis Functional enrichment analysis (TANGO)(TANGO) 714 enriched for GO ldquobiological 714 enriched for GO ldquobiological
processrdquoprocessrdquo 693 enriched for GO693 enriched for GO ldquocellular ldquocellular
compartmentrdquocompartmentrdquo 463 of known complexes enriched in 463 of known complexes enriched in
at least one BPMat least one BPM
Pivots tend to be Pivots tend to be multifunctionalmultifunctional
No of No of proteinsproteins
PivotsPivots ExpectedExpected SignificanceSignificance
GO complexesGO complexes 206206 2121 435435 pp=127=1271010-9-9
KEGG pathwaysKEGG pathways 7171 66 150150 pp=368=368 1010-3-3
Curated multi-Curated multi-complexed proteins complexed proteins (Krause et al 2006)(Krause et al 2006) 3939 88 055055 pp=749=749 1010-9-9
Pivot proteins are essentialPivot proteins are essential
Q Are proteins active in multiple partially Q Are proteins active in multiple partially redundant pathways more essentialredundant pathways more essential72124 (58) pivots essential (72124 (58) pivots essential (pp=142middot10=142middot10-23-23))Enrichment is not explained by node degrees Enrichment is not explained by node degrees ((pplt10lt10-5-5))Essential pivots closer in function to their BPMs Essential pivots closer in function to their BPMs than nonessential pivotsthan nonessential pivotsPivots significantly retained during evolution Pivots significantly retained during evolution (p=979middot10(p=979middot10-9-9))
SWR1 Ino80
SWR1Ino80 ExampleSWR1Ino80 ExamplePivots
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
Genetic interactions (GI)Genetic interactions (GI)
ldquoldquoAggravatingrdquo interactionsAggravatingrdquo interactions
The observed phenotype is The observed phenotype is worseworse than what we would ldquoexpectrdquothan what we would ldquoexpectrdquo
Synthetic lethalitySynthetic lethality joint deletion of joint deletion of two nonessential genes two nonessential genes lethal lethal phenotypephenotype
Synthetic sicknessSynthetic sickness joint deletion of joint deletion of two nonessential genes two nonessential genes slow slow growthgrowth
GI availabilityGI availability
Systematically mapped by SGA and Systematically mapped by SGA and dSLAMdSLAM
Available for asymp200 gene queries and Available for asymp200 gene queries and asymp4500 targets in asymp4500 targets in S cerevisiaeS cerevisiae
GI network 13632 interactions GI network 13632 interactions 2682 genes2682 genes
Physical interactions (PI)Physical interactions (PI)
Protein-protein interactionsProtein-protein interactions Y2HY2H TAPTAP
Protein-DNA interactionsProtein-DNA interactions ChipChip22 experiments experiments
PI network 68172 interactions 6184 PI network 68172 interactions 6184 proteinsproteins
GI Analysis SpectraGI Analysis Spectra
Genetically interacting homologs tend to Genetically interacting homologs tend to exhibit ldquocompensationrdquo on the expression exhibit ldquocompensationrdquo on the expression levellevel R Kafri R Kafri et alet al 2005 2005
Genetically interacting proteins frequently Genetically interacting proteins frequently have a similar foldhave a similar fold O Dror O Dror et alet al 2007 2007
Genetic interactions can be used to Genetic interactions can be used to delineate regulatory pathwaysdelineate regulatory pathways RP Onge RP Onge et alet al 2007 2007
Joint analysis of GI and PIJoint analysis of GI and PI
Motivation Motivation Identifying pathwaysIdentifying pathways Connect pathways with phenotypesConnect pathways with phenotypes Understand system features (robustness Understand system features (robustness
essentialityhellip)essentialityhellip)
Initial analysisInitial analysis Proteins close in the GI network slightly more Proteins close in the GI network slightly more
likely to physically interact (Tong et al 2004) likely to physically interact (Tong et al 2004) PI hubs likely to be GI hubs (Ozier et al PI hubs likely to be GI hubs (Ozier et al
2003)2003)
BPMsBPMs
Kelley and Ideker 2005 Kelley and Ideker 2005 modeling ldquoexplanationsrdquo for GIsmodeling ldquoexplanationsrdquo for GIs
Within-pathway model Between-pathway model
GI
PI
Kelley amp Ideker conclusionsKelley amp Ideker conclusions
40 of the GIs can be explained by physical 40 of the GIs can be explained by physical modelsmodels
Between-pathway models explain x 35 times Between-pathway models explain x 35 times more GIs than within-pathway modelsmore GIs than within-pathway models
Models capable of predicting protein functionsModels capable of predicting protein functions
BPM rationaleBPM rationale
Between pathway Between pathway models suggest models suggest redundancyredundancyAlternative pathsAlternative paths Independent KOs Independent KOs
little effectlittle effect Joint KOs in both Joint KOs in both
severe effectsevere effect
Viable phenotypeViable phenotype
Lethal phenotype
BPM HuntingBPM Hunting
Look for pairs of pathways with GI Look for pairs of pathways with GI evidence for bufferingevidence for buffering
A pathway ndash a A pathway ndash a connectedconnected subnetwork in subnetwork in of PIsof PIs
Model scoring based on the density of GIsModel scoring based on the density of GIs
Log-likelihood scores accounting for GI Log-likelihood scores accounting for GI degreesdegrees
1 2
1 2
( )1 2 BPM1 2
1 2 null ( )
( ) (1 )(1 ( ))( | )( ) log log
( | ) ( ) (1 )(1 ( ))a b V V
genetica b a ba b V V
I a b I a bP V V MS V V
P V V M r I a b r I a b
AlgorithmicsAlgorithmics
High-scoring seedsHigh-scoring seeds Finding heavy bicliques Finding heavy bicliques
with connectivity with connectivity constraintsconstraints
Seed optimizationSeed optimization Greedy search Greedy search
maintaining maintaining connectivityconnectivity
Significance filteringSignificance filtering
Adding ldquopivot proteinsrdquo to Adding ldquopivot proteinsrdquo to BPMsBPMs
Pathway bifurcation Redundant sub-complexes
Pivot proteinsPivot proteins
Physically connected to both pathways Physically connected to both pathways in the BPMin the BPM
Connection significant given the Connection significant given the general network degreesgeneral network degrees
Analysis flowAnalysis flow
13632 genetic interactions
68172 physical interactions
140 models
124 pivot proteins in 40 models
BPM analysis
Pivot extraction
Physiological characteristics
bullEssentialitybullPhenotypesbullProtein AbundancebullCodon adaptation indexbullmRNA half-lifebullPhoshorylation
BPMs as functional modulesBPMs as functional modules
Functional enrichment analysis Functional enrichment analysis (TANGO)(TANGO) 714 enriched for GO ldquobiological 714 enriched for GO ldquobiological
processrdquoprocessrdquo 693 enriched for GO693 enriched for GO ldquocellular ldquocellular
compartmentrdquocompartmentrdquo 463 of known complexes enriched in 463 of known complexes enriched in
at least one BPMat least one BPM
Pivots tend to be Pivots tend to be multifunctionalmultifunctional
No of No of proteinsproteins
PivotsPivots ExpectedExpected SignificanceSignificance
GO complexesGO complexes 206206 2121 435435 pp=127=1271010-9-9
KEGG pathwaysKEGG pathways 7171 66 150150 pp=368=368 1010-3-3
Curated multi-Curated multi-complexed proteins complexed proteins (Krause et al 2006)(Krause et al 2006) 3939 88 055055 pp=749=749 1010-9-9
Pivot proteins are essentialPivot proteins are essential
Q Are proteins active in multiple partially Q Are proteins active in multiple partially redundant pathways more essentialredundant pathways more essential72124 (58) pivots essential (72124 (58) pivots essential (pp=142middot10=142middot10-23-23))Enrichment is not explained by node degrees Enrichment is not explained by node degrees ((pplt10lt10-5-5))Essential pivots closer in function to their BPMs Essential pivots closer in function to their BPMs than nonessential pivotsthan nonessential pivotsPivots significantly retained during evolution Pivots significantly retained during evolution (p=979middot10(p=979middot10-9-9))
SWR1 Ino80
SWR1Ino80 ExampleSWR1Ino80 ExamplePivots
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
GI availabilityGI availability
Systematically mapped by SGA and Systematically mapped by SGA and dSLAMdSLAM
Available for asymp200 gene queries and Available for asymp200 gene queries and asymp4500 targets in asymp4500 targets in S cerevisiaeS cerevisiae
GI network 13632 interactions GI network 13632 interactions 2682 genes2682 genes
Physical interactions (PI)Physical interactions (PI)
Protein-protein interactionsProtein-protein interactions Y2HY2H TAPTAP
Protein-DNA interactionsProtein-DNA interactions ChipChip22 experiments experiments
PI network 68172 interactions 6184 PI network 68172 interactions 6184 proteinsproteins
GI Analysis SpectraGI Analysis Spectra
Genetically interacting homologs tend to Genetically interacting homologs tend to exhibit ldquocompensationrdquo on the expression exhibit ldquocompensationrdquo on the expression levellevel R Kafri R Kafri et alet al 2005 2005
Genetically interacting proteins frequently Genetically interacting proteins frequently have a similar foldhave a similar fold O Dror O Dror et alet al 2007 2007
Genetic interactions can be used to Genetic interactions can be used to delineate regulatory pathwaysdelineate regulatory pathways RP Onge RP Onge et alet al 2007 2007
Joint analysis of GI and PIJoint analysis of GI and PI
Motivation Motivation Identifying pathwaysIdentifying pathways Connect pathways with phenotypesConnect pathways with phenotypes Understand system features (robustness Understand system features (robustness
essentialityhellip)essentialityhellip)
Initial analysisInitial analysis Proteins close in the GI network slightly more Proteins close in the GI network slightly more
likely to physically interact (Tong et al 2004) likely to physically interact (Tong et al 2004) PI hubs likely to be GI hubs (Ozier et al PI hubs likely to be GI hubs (Ozier et al
2003)2003)
BPMsBPMs
Kelley and Ideker 2005 Kelley and Ideker 2005 modeling ldquoexplanationsrdquo for GIsmodeling ldquoexplanationsrdquo for GIs
Within-pathway model Between-pathway model
GI
PI
Kelley amp Ideker conclusionsKelley amp Ideker conclusions
40 of the GIs can be explained by physical 40 of the GIs can be explained by physical modelsmodels
Between-pathway models explain x 35 times Between-pathway models explain x 35 times more GIs than within-pathway modelsmore GIs than within-pathway models
Models capable of predicting protein functionsModels capable of predicting protein functions
BPM rationaleBPM rationale
Between pathway Between pathway models suggest models suggest redundancyredundancyAlternative pathsAlternative paths Independent KOs Independent KOs
little effectlittle effect Joint KOs in both Joint KOs in both
severe effectsevere effect
Viable phenotypeViable phenotype
Lethal phenotype
BPM HuntingBPM Hunting
Look for pairs of pathways with GI Look for pairs of pathways with GI evidence for bufferingevidence for buffering
A pathway ndash a A pathway ndash a connectedconnected subnetwork in subnetwork in of PIsof PIs
Model scoring based on the density of GIsModel scoring based on the density of GIs
Log-likelihood scores accounting for GI Log-likelihood scores accounting for GI degreesdegrees
1 2
1 2
( )1 2 BPM1 2
1 2 null ( )
( ) (1 )(1 ( ))( | )( ) log log
( | ) ( ) (1 )(1 ( ))a b V V
genetica b a ba b V V
I a b I a bP V V MS V V
P V V M r I a b r I a b
AlgorithmicsAlgorithmics
High-scoring seedsHigh-scoring seeds Finding heavy bicliques Finding heavy bicliques
with connectivity with connectivity constraintsconstraints
Seed optimizationSeed optimization Greedy search Greedy search
maintaining maintaining connectivityconnectivity
Significance filteringSignificance filtering
Adding ldquopivot proteinsrdquo to Adding ldquopivot proteinsrdquo to BPMsBPMs
Pathway bifurcation Redundant sub-complexes
Pivot proteinsPivot proteins
Physically connected to both pathways Physically connected to both pathways in the BPMin the BPM
Connection significant given the Connection significant given the general network degreesgeneral network degrees
Analysis flowAnalysis flow
13632 genetic interactions
68172 physical interactions
140 models
124 pivot proteins in 40 models
BPM analysis
Pivot extraction
Physiological characteristics
bullEssentialitybullPhenotypesbullProtein AbundancebullCodon adaptation indexbullmRNA half-lifebullPhoshorylation
BPMs as functional modulesBPMs as functional modules
Functional enrichment analysis Functional enrichment analysis (TANGO)(TANGO) 714 enriched for GO ldquobiological 714 enriched for GO ldquobiological
processrdquoprocessrdquo 693 enriched for GO693 enriched for GO ldquocellular ldquocellular
compartmentrdquocompartmentrdquo 463 of known complexes enriched in 463 of known complexes enriched in
at least one BPMat least one BPM
Pivots tend to be Pivots tend to be multifunctionalmultifunctional
No of No of proteinsproteins
PivotsPivots ExpectedExpected SignificanceSignificance
GO complexesGO complexes 206206 2121 435435 pp=127=1271010-9-9
KEGG pathwaysKEGG pathways 7171 66 150150 pp=368=368 1010-3-3
Curated multi-Curated multi-complexed proteins complexed proteins (Krause et al 2006)(Krause et al 2006) 3939 88 055055 pp=749=749 1010-9-9
Pivot proteins are essentialPivot proteins are essential
Q Are proteins active in multiple partially Q Are proteins active in multiple partially redundant pathways more essentialredundant pathways more essential72124 (58) pivots essential (72124 (58) pivots essential (pp=142middot10=142middot10-23-23))Enrichment is not explained by node degrees Enrichment is not explained by node degrees ((pplt10lt10-5-5))Essential pivots closer in function to their BPMs Essential pivots closer in function to their BPMs than nonessential pivotsthan nonessential pivotsPivots significantly retained during evolution Pivots significantly retained during evolution (p=979middot10(p=979middot10-9-9))
SWR1 Ino80
SWR1Ino80 ExampleSWR1Ino80 ExamplePivots
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
Physical interactions (PI)Physical interactions (PI)
Protein-protein interactionsProtein-protein interactions Y2HY2H TAPTAP
Protein-DNA interactionsProtein-DNA interactions ChipChip22 experiments experiments
PI network 68172 interactions 6184 PI network 68172 interactions 6184 proteinsproteins
GI Analysis SpectraGI Analysis Spectra
Genetically interacting homologs tend to Genetically interacting homologs tend to exhibit ldquocompensationrdquo on the expression exhibit ldquocompensationrdquo on the expression levellevel R Kafri R Kafri et alet al 2005 2005
Genetically interacting proteins frequently Genetically interacting proteins frequently have a similar foldhave a similar fold O Dror O Dror et alet al 2007 2007
Genetic interactions can be used to Genetic interactions can be used to delineate regulatory pathwaysdelineate regulatory pathways RP Onge RP Onge et alet al 2007 2007
Joint analysis of GI and PIJoint analysis of GI and PI
Motivation Motivation Identifying pathwaysIdentifying pathways Connect pathways with phenotypesConnect pathways with phenotypes Understand system features (robustness Understand system features (robustness
essentialityhellip)essentialityhellip)
Initial analysisInitial analysis Proteins close in the GI network slightly more Proteins close in the GI network slightly more
likely to physically interact (Tong et al 2004) likely to physically interact (Tong et al 2004) PI hubs likely to be GI hubs (Ozier et al PI hubs likely to be GI hubs (Ozier et al
2003)2003)
BPMsBPMs
Kelley and Ideker 2005 Kelley and Ideker 2005 modeling ldquoexplanationsrdquo for GIsmodeling ldquoexplanationsrdquo for GIs
Within-pathway model Between-pathway model
GI
PI
Kelley amp Ideker conclusionsKelley amp Ideker conclusions
40 of the GIs can be explained by physical 40 of the GIs can be explained by physical modelsmodels
Between-pathway models explain x 35 times Between-pathway models explain x 35 times more GIs than within-pathway modelsmore GIs than within-pathway models
Models capable of predicting protein functionsModels capable of predicting protein functions
BPM rationaleBPM rationale
Between pathway Between pathway models suggest models suggest redundancyredundancyAlternative pathsAlternative paths Independent KOs Independent KOs
little effectlittle effect Joint KOs in both Joint KOs in both
severe effectsevere effect
Viable phenotypeViable phenotype
Lethal phenotype
BPM HuntingBPM Hunting
Look for pairs of pathways with GI Look for pairs of pathways with GI evidence for bufferingevidence for buffering
A pathway ndash a A pathway ndash a connectedconnected subnetwork in subnetwork in of PIsof PIs
Model scoring based on the density of GIsModel scoring based on the density of GIs
Log-likelihood scores accounting for GI Log-likelihood scores accounting for GI degreesdegrees
1 2
1 2
( )1 2 BPM1 2
1 2 null ( )
( ) (1 )(1 ( ))( | )( ) log log
( | ) ( ) (1 )(1 ( ))a b V V
genetica b a ba b V V
I a b I a bP V V MS V V
P V V M r I a b r I a b
AlgorithmicsAlgorithmics
High-scoring seedsHigh-scoring seeds Finding heavy bicliques Finding heavy bicliques
with connectivity with connectivity constraintsconstraints
Seed optimizationSeed optimization Greedy search Greedy search
maintaining maintaining connectivityconnectivity
Significance filteringSignificance filtering
Adding ldquopivot proteinsrdquo to Adding ldquopivot proteinsrdquo to BPMsBPMs
Pathway bifurcation Redundant sub-complexes
Pivot proteinsPivot proteins
Physically connected to both pathways Physically connected to both pathways in the BPMin the BPM
Connection significant given the Connection significant given the general network degreesgeneral network degrees
Analysis flowAnalysis flow
13632 genetic interactions
68172 physical interactions
140 models
124 pivot proteins in 40 models
BPM analysis
Pivot extraction
Physiological characteristics
bullEssentialitybullPhenotypesbullProtein AbundancebullCodon adaptation indexbullmRNA half-lifebullPhoshorylation
BPMs as functional modulesBPMs as functional modules
Functional enrichment analysis Functional enrichment analysis (TANGO)(TANGO) 714 enriched for GO ldquobiological 714 enriched for GO ldquobiological
processrdquoprocessrdquo 693 enriched for GO693 enriched for GO ldquocellular ldquocellular
compartmentrdquocompartmentrdquo 463 of known complexes enriched in 463 of known complexes enriched in
at least one BPMat least one BPM
Pivots tend to be Pivots tend to be multifunctionalmultifunctional
No of No of proteinsproteins
PivotsPivots ExpectedExpected SignificanceSignificance
GO complexesGO complexes 206206 2121 435435 pp=127=1271010-9-9
KEGG pathwaysKEGG pathways 7171 66 150150 pp=368=368 1010-3-3
Curated multi-Curated multi-complexed proteins complexed proteins (Krause et al 2006)(Krause et al 2006) 3939 88 055055 pp=749=749 1010-9-9
Pivot proteins are essentialPivot proteins are essential
Q Are proteins active in multiple partially Q Are proteins active in multiple partially redundant pathways more essentialredundant pathways more essential72124 (58) pivots essential (72124 (58) pivots essential (pp=142middot10=142middot10-23-23))Enrichment is not explained by node degrees Enrichment is not explained by node degrees ((pplt10lt10-5-5))Essential pivots closer in function to their BPMs Essential pivots closer in function to their BPMs than nonessential pivotsthan nonessential pivotsPivots significantly retained during evolution Pivots significantly retained during evolution (p=979middot10(p=979middot10-9-9))
SWR1 Ino80
SWR1Ino80 ExampleSWR1Ino80 ExamplePivots
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
GI Analysis SpectraGI Analysis Spectra
Genetically interacting homologs tend to Genetically interacting homologs tend to exhibit ldquocompensationrdquo on the expression exhibit ldquocompensationrdquo on the expression levellevel R Kafri R Kafri et alet al 2005 2005
Genetically interacting proteins frequently Genetically interacting proteins frequently have a similar foldhave a similar fold O Dror O Dror et alet al 2007 2007
Genetic interactions can be used to Genetic interactions can be used to delineate regulatory pathwaysdelineate regulatory pathways RP Onge RP Onge et alet al 2007 2007
Joint analysis of GI and PIJoint analysis of GI and PI
Motivation Motivation Identifying pathwaysIdentifying pathways Connect pathways with phenotypesConnect pathways with phenotypes Understand system features (robustness Understand system features (robustness
essentialityhellip)essentialityhellip)
Initial analysisInitial analysis Proteins close in the GI network slightly more Proteins close in the GI network slightly more
likely to physically interact (Tong et al 2004) likely to physically interact (Tong et al 2004) PI hubs likely to be GI hubs (Ozier et al PI hubs likely to be GI hubs (Ozier et al
2003)2003)
BPMsBPMs
Kelley and Ideker 2005 Kelley and Ideker 2005 modeling ldquoexplanationsrdquo for GIsmodeling ldquoexplanationsrdquo for GIs
Within-pathway model Between-pathway model
GI
PI
Kelley amp Ideker conclusionsKelley amp Ideker conclusions
40 of the GIs can be explained by physical 40 of the GIs can be explained by physical modelsmodels
Between-pathway models explain x 35 times Between-pathway models explain x 35 times more GIs than within-pathway modelsmore GIs than within-pathway models
Models capable of predicting protein functionsModels capable of predicting protein functions
BPM rationaleBPM rationale
Between pathway Between pathway models suggest models suggest redundancyredundancyAlternative pathsAlternative paths Independent KOs Independent KOs
little effectlittle effect Joint KOs in both Joint KOs in both
severe effectsevere effect
Viable phenotypeViable phenotype
Lethal phenotype
BPM HuntingBPM Hunting
Look for pairs of pathways with GI Look for pairs of pathways with GI evidence for bufferingevidence for buffering
A pathway ndash a A pathway ndash a connectedconnected subnetwork in subnetwork in of PIsof PIs
Model scoring based on the density of GIsModel scoring based on the density of GIs
Log-likelihood scores accounting for GI Log-likelihood scores accounting for GI degreesdegrees
1 2
1 2
( )1 2 BPM1 2
1 2 null ( )
( ) (1 )(1 ( ))( | )( ) log log
( | ) ( ) (1 )(1 ( ))a b V V
genetica b a ba b V V
I a b I a bP V V MS V V
P V V M r I a b r I a b
AlgorithmicsAlgorithmics
High-scoring seedsHigh-scoring seeds Finding heavy bicliques Finding heavy bicliques
with connectivity with connectivity constraintsconstraints
Seed optimizationSeed optimization Greedy search Greedy search
maintaining maintaining connectivityconnectivity
Significance filteringSignificance filtering
Adding ldquopivot proteinsrdquo to Adding ldquopivot proteinsrdquo to BPMsBPMs
Pathway bifurcation Redundant sub-complexes
Pivot proteinsPivot proteins
Physically connected to both pathways Physically connected to both pathways in the BPMin the BPM
Connection significant given the Connection significant given the general network degreesgeneral network degrees
Analysis flowAnalysis flow
13632 genetic interactions
68172 physical interactions
140 models
124 pivot proteins in 40 models
BPM analysis
Pivot extraction
Physiological characteristics
bullEssentialitybullPhenotypesbullProtein AbundancebullCodon adaptation indexbullmRNA half-lifebullPhoshorylation
BPMs as functional modulesBPMs as functional modules
Functional enrichment analysis Functional enrichment analysis (TANGO)(TANGO) 714 enriched for GO ldquobiological 714 enriched for GO ldquobiological
processrdquoprocessrdquo 693 enriched for GO693 enriched for GO ldquocellular ldquocellular
compartmentrdquocompartmentrdquo 463 of known complexes enriched in 463 of known complexes enriched in
at least one BPMat least one BPM
Pivots tend to be Pivots tend to be multifunctionalmultifunctional
No of No of proteinsproteins
PivotsPivots ExpectedExpected SignificanceSignificance
GO complexesGO complexes 206206 2121 435435 pp=127=1271010-9-9
KEGG pathwaysKEGG pathways 7171 66 150150 pp=368=368 1010-3-3
Curated multi-Curated multi-complexed proteins complexed proteins (Krause et al 2006)(Krause et al 2006) 3939 88 055055 pp=749=749 1010-9-9
Pivot proteins are essentialPivot proteins are essential
Q Are proteins active in multiple partially Q Are proteins active in multiple partially redundant pathways more essentialredundant pathways more essential72124 (58) pivots essential (72124 (58) pivots essential (pp=142middot10=142middot10-23-23))Enrichment is not explained by node degrees Enrichment is not explained by node degrees ((pplt10lt10-5-5))Essential pivots closer in function to their BPMs Essential pivots closer in function to their BPMs than nonessential pivotsthan nonessential pivotsPivots significantly retained during evolution Pivots significantly retained during evolution (p=979middot10(p=979middot10-9-9))
SWR1 Ino80
SWR1Ino80 ExampleSWR1Ino80 ExamplePivots
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
Joint analysis of GI and PIJoint analysis of GI and PI
Motivation Motivation Identifying pathwaysIdentifying pathways Connect pathways with phenotypesConnect pathways with phenotypes Understand system features (robustness Understand system features (robustness
essentialityhellip)essentialityhellip)
Initial analysisInitial analysis Proteins close in the GI network slightly more Proteins close in the GI network slightly more
likely to physically interact (Tong et al 2004) likely to physically interact (Tong et al 2004) PI hubs likely to be GI hubs (Ozier et al PI hubs likely to be GI hubs (Ozier et al
2003)2003)
BPMsBPMs
Kelley and Ideker 2005 Kelley and Ideker 2005 modeling ldquoexplanationsrdquo for GIsmodeling ldquoexplanationsrdquo for GIs
Within-pathway model Between-pathway model
GI
PI
Kelley amp Ideker conclusionsKelley amp Ideker conclusions
40 of the GIs can be explained by physical 40 of the GIs can be explained by physical modelsmodels
Between-pathway models explain x 35 times Between-pathway models explain x 35 times more GIs than within-pathway modelsmore GIs than within-pathway models
Models capable of predicting protein functionsModels capable of predicting protein functions
BPM rationaleBPM rationale
Between pathway Between pathway models suggest models suggest redundancyredundancyAlternative pathsAlternative paths Independent KOs Independent KOs
little effectlittle effect Joint KOs in both Joint KOs in both
severe effectsevere effect
Viable phenotypeViable phenotype
Lethal phenotype
BPM HuntingBPM Hunting
Look for pairs of pathways with GI Look for pairs of pathways with GI evidence for bufferingevidence for buffering
A pathway ndash a A pathway ndash a connectedconnected subnetwork in subnetwork in of PIsof PIs
Model scoring based on the density of GIsModel scoring based on the density of GIs
Log-likelihood scores accounting for GI Log-likelihood scores accounting for GI degreesdegrees
1 2
1 2
( )1 2 BPM1 2
1 2 null ( )
( ) (1 )(1 ( ))( | )( ) log log
( | ) ( ) (1 )(1 ( ))a b V V
genetica b a ba b V V
I a b I a bP V V MS V V
P V V M r I a b r I a b
AlgorithmicsAlgorithmics
High-scoring seedsHigh-scoring seeds Finding heavy bicliques Finding heavy bicliques
with connectivity with connectivity constraintsconstraints
Seed optimizationSeed optimization Greedy search Greedy search
maintaining maintaining connectivityconnectivity
Significance filteringSignificance filtering
Adding ldquopivot proteinsrdquo to Adding ldquopivot proteinsrdquo to BPMsBPMs
Pathway bifurcation Redundant sub-complexes
Pivot proteinsPivot proteins
Physically connected to both pathways Physically connected to both pathways in the BPMin the BPM
Connection significant given the Connection significant given the general network degreesgeneral network degrees
Analysis flowAnalysis flow
13632 genetic interactions
68172 physical interactions
140 models
124 pivot proteins in 40 models
BPM analysis
Pivot extraction
Physiological characteristics
bullEssentialitybullPhenotypesbullProtein AbundancebullCodon adaptation indexbullmRNA half-lifebullPhoshorylation
BPMs as functional modulesBPMs as functional modules
Functional enrichment analysis Functional enrichment analysis (TANGO)(TANGO) 714 enriched for GO ldquobiological 714 enriched for GO ldquobiological
processrdquoprocessrdquo 693 enriched for GO693 enriched for GO ldquocellular ldquocellular
compartmentrdquocompartmentrdquo 463 of known complexes enriched in 463 of known complexes enriched in
at least one BPMat least one BPM
Pivots tend to be Pivots tend to be multifunctionalmultifunctional
No of No of proteinsproteins
PivotsPivots ExpectedExpected SignificanceSignificance
GO complexesGO complexes 206206 2121 435435 pp=127=1271010-9-9
KEGG pathwaysKEGG pathways 7171 66 150150 pp=368=368 1010-3-3
Curated multi-Curated multi-complexed proteins complexed proteins (Krause et al 2006)(Krause et al 2006) 3939 88 055055 pp=749=749 1010-9-9
Pivot proteins are essentialPivot proteins are essential
Q Are proteins active in multiple partially Q Are proteins active in multiple partially redundant pathways more essentialredundant pathways more essential72124 (58) pivots essential (72124 (58) pivots essential (pp=142middot10=142middot10-23-23))Enrichment is not explained by node degrees Enrichment is not explained by node degrees ((pplt10lt10-5-5))Essential pivots closer in function to their BPMs Essential pivots closer in function to their BPMs than nonessential pivotsthan nonessential pivotsPivots significantly retained during evolution Pivots significantly retained during evolution (p=979middot10(p=979middot10-9-9))
SWR1 Ino80
SWR1Ino80 ExampleSWR1Ino80 ExamplePivots
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
BPMsBPMs
Kelley and Ideker 2005 Kelley and Ideker 2005 modeling ldquoexplanationsrdquo for GIsmodeling ldquoexplanationsrdquo for GIs
Within-pathway model Between-pathway model
GI
PI
Kelley amp Ideker conclusionsKelley amp Ideker conclusions
40 of the GIs can be explained by physical 40 of the GIs can be explained by physical modelsmodels
Between-pathway models explain x 35 times Between-pathway models explain x 35 times more GIs than within-pathway modelsmore GIs than within-pathway models
Models capable of predicting protein functionsModels capable of predicting protein functions
BPM rationaleBPM rationale
Between pathway Between pathway models suggest models suggest redundancyredundancyAlternative pathsAlternative paths Independent KOs Independent KOs
little effectlittle effect Joint KOs in both Joint KOs in both
severe effectsevere effect
Viable phenotypeViable phenotype
Lethal phenotype
BPM HuntingBPM Hunting
Look for pairs of pathways with GI Look for pairs of pathways with GI evidence for bufferingevidence for buffering
A pathway ndash a A pathway ndash a connectedconnected subnetwork in subnetwork in of PIsof PIs
Model scoring based on the density of GIsModel scoring based on the density of GIs
Log-likelihood scores accounting for GI Log-likelihood scores accounting for GI degreesdegrees
1 2
1 2
( )1 2 BPM1 2
1 2 null ( )
( ) (1 )(1 ( ))( | )( ) log log
( | ) ( ) (1 )(1 ( ))a b V V
genetica b a ba b V V
I a b I a bP V V MS V V
P V V M r I a b r I a b
AlgorithmicsAlgorithmics
High-scoring seedsHigh-scoring seeds Finding heavy bicliques Finding heavy bicliques
with connectivity with connectivity constraintsconstraints
Seed optimizationSeed optimization Greedy search Greedy search
maintaining maintaining connectivityconnectivity
Significance filteringSignificance filtering
Adding ldquopivot proteinsrdquo to Adding ldquopivot proteinsrdquo to BPMsBPMs
Pathway bifurcation Redundant sub-complexes
Pivot proteinsPivot proteins
Physically connected to both pathways Physically connected to both pathways in the BPMin the BPM
Connection significant given the Connection significant given the general network degreesgeneral network degrees
Analysis flowAnalysis flow
13632 genetic interactions
68172 physical interactions
140 models
124 pivot proteins in 40 models
BPM analysis
Pivot extraction
Physiological characteristics
bullEssentialitybullPhenotypesbullProtein AbundancebullCodon adaptation indexbullmRNA half-lifebullPhoshorylation
BPMs as functional modulesBPMs as functional modules
Functional enrichment analysis Functional enrichment analysis (TANGO)(TANGO) 714 enriched for GO ldquobiological 714 enriched for GO ldquobiological
processrdquoprocessrdquo 693 enriched for GO693 enriched for GO ldquocellular ldquocellular
compartmentrdquocompartmentrdquo 463 of known complexes enriched in 463 of known complexes enriched in
at least one BPMat least one BPM
Pivots tend to be Pivots tend to be multifunctionalmultifunctional
No of No of proteinsproteins
PivotsPivots ExpectedExpected SignificanceSignificance
GO complexesGO complexes 206206 2121 435435 pp=127=1271010-9-9
KEGG pathwaysKEGG pathways 7171 66 150150 pp=368=368 1010-3-3
Curated multi-Curated multi-complexed proteins complexed proteins (Krause et al 2006)(Krause et al 2006) 3939 88 055055 pp=749=749 1010-9-9
Pivot proteins are essentialPivot proteins are essential
Q Are proteins active in multiple partially Q Are proteins active in multiple partially redundant pathways more essentialredundant pathways more essential72124 (58) pivots essential (72124 (58) pivots essential (pp=142middot10=142middot10-23-23))Enrichment is not explained by node degrees Enrichment is not explained by node degrees ((pplt10lt10-5-5))Essential pivots closer in function to their BPMs Essential pivots closer in function to their BPMs than nonessential pivotsthan nonessential pivotsPivots significantly retained during evolution Pivots significantly retained during evolution (p=979middot10(p=979middot10-9-9))
SWR1 Ino80
SWR1Ino80 ExampleSWR1Ino80 ExamplePivots
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
Kelley amp Ideker conclusionsKelley amp Ideker conclusions
40 of the GIs can be explained by physical 40 of the GIs can be explained by physical modelsmodels
Between-pathway models explain x 35 times Between-pathway models explain x 35 times more GIs than within-pathway modelsmore GIs than within-pathway models
Models capable of predicting protein functionsModels capable of predicting protein functions
BPM rationaleBPM rationale
Between pathway Between pathway models suggest models suggest redundancyredundancyAlternative pathsAlternative paths Independent KOs Independent KOs
little effectlittle effect Joint KOs in both Joint KOs in both
severe effectsevere effect
Viable phenotypeViable phenotype
Lethal phenotype
BPM HuntingBPM Hunting
Look for pairs of pathways with GI Look for pairs of pathways with GI evidence for bufferingevidence for buffering
A pathway ndash a A pathway ndash a connectedconnected subnetwork in subnetwork in of PIsof PIs
Model scoring based on the density of GIsModel scoring based on the density of GIs
Log-likelihood scores accounting for GI Log-likelihood scores accounting for GI degreesdegrees
1 2
1 2
( )1 2 BPM1 2
1 2 null ( )
( ) (1 )(1 ( ))( | )( ) log log
( | ) ( ) (1 )(1 ( ))a b V V
genetica b a ba b V V
I a b I a bP V V MS V V
P V V M r I a b r I a b
AlgorithmicsAlgorithmics
High-scoring seedsHigh-scoring seeds Finding heavy bicliques Finding heavy bicliques
with connectivity with connectivity constraintsconstraints
Seed optimizationSeed optimization Greedy search Greedy search
maintaining maintaining connectivityconnectivity
Significance filteringSignificance filtering
Adding ldquopivot proteinsrdquo to Adding ldquopivot proteinsrdquo to BPMsBPMs
Pathway bifurcation Redundant sub-complexes
Pivot proteinsPivot proteins
Physically connected to both pathways Physically connected to both pathways in the BPMin the BPM
Connection significant given the Connection significant given the general network degreesgeneral network degrees
Analysis flowAnalysis flow
13632 genetic interactions
68172 physical interactions
140 models
124 pivot proteins in 40 models
BPM analysis
Pivot extraction
Physiological characteristics
bullEssentialitybullPhenotypesbullProtein AbundancebullCodon adaptation indexbullmRNA half-lifebullPhoshorylation
BPMs as functional modulesBPMs as functional modules
Functional enrichment analysis Functional enrichment analysis (TANGO)(TANGO) 714 enriched for GO ldquobiological 714 enriched for GO ldquobiological
processrdquoprocessrdquo 693 enriched for GO693 enriched for GO ldquocellular ldquocellular
compartmentrdquocompartmentrdquo 463 of known complexes enriched in 463 of known complexes enriched in
at least one BPMat least one BPM
Pivots tend to be Pivots tend to be multifunctionalmultifunctional
No of No of proteinsproteins
PivotsPivots ExpectedExpected SignificanceSignificance
GO complexesGO complexes 206206 2121 435435 pp=127=1271010-9-9
KEGG pathwaysKEGG pathways 7171 66 150150 pp=368=368 1010-3-3
Curated multi-Curated multi-complexed proteins complexed proteins (Krause et al 2006)(Krause et al 2006) 3939 88 055055 pp=749=749 1010-9-9
Pivot proteins are essentialPivot proteins are essential
Q Are proteins active in multiple partially Q Are proteins active in multiple partially redundant pathways more essentialredundant pathways more essential72124 (58) pivots essential (72124 (58) pivots essential (pp=142middot10=142middot10-23-23))Enrichment is not explained by node degrees Enrichment is not explained by node degrees ((pplt10lt10-5-5))Essential pivots closer in function to their BPMs Essential pivots closer in function to their BPMs than nonessential pivotsthan nonessential pivotsPivots significantly retained during evolution Pivots significantly retained during evolution (p=979middot10(p=979middot10-9-9))
SWR1 Ino80
SWR1Ino80 ExampleSWR1Ino80 ExamplePivots
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
BPM rationaleBPM rationale
Between pathway Between pathway models suggest models suggest redundancyredundancyAlternative pathsAlternative paths Independent KOs Independent KOs
little effectlittle effect Joint KOs in both Joint KOs in both
severe effectsevere effect
Viable phenotypeViable phenotype
Lethal phenotype
BPM HuntingBPM Hunting
Look for pairs of pathways with GI Look for pairs of pathways with GI evidence for bufferingevidence for buffering
A pathway ndash a A pathway ndash a connectedconnected subnetwork in subnetwork in of PIsof PIs
Model scoring based on the density of GIsModel scoring based on the density of GIs
Log-likelihood scores accounting for GI Log-likelihood scores accounting for GI degreesdegrees
1 2
1 2
( )1 2 BPM1 2
1 2 null ( )
( ) (1 )(1 ( ))( | )( ) log log
( | ) ( ) (1 )(1 ( ))a b V V
genetica b a ba b V V
I a b I a bP V V MS V V
P V V M r I a b r I a b
AlgorithmicsAlgorithmics
High-scoring seedsHigh-scoring seeds Finding heavy bicliques Finding heavy bicliques
with connectivity with connectivity constraintsconstraints
Seed optimizationSeed optimization Greedy search Greedy search
maintaining maintaining connectivityconnectivity
Significance filteringSignificance filtering
Adding ldquopivot proteinsrdquo to Adding ldquopivot proteinsrdquo to BPMsBPMs
Pathway bifurcation Redundant sub-complexes
Pivot proteinsPivot proteins
Physically connected to both pathways Physically connected to both pathways in the BPMin the BPM
Connection significant given the Connection significant given the general network degreesgeneral network degrees
Analysis flowAnalysis flow
13632 genetic interactions
68172 physical interactions
140 models
124 pivot proteins in 40 models
BPM analysis
Pivot extraction
Physiological characteristics
bullEssentialitybullPhenotypesbullProtein AbundancebullCodon adaptation indexbullmRNA half-lifebullPhoshorylation
BPMs as functional modulesBPMs as functional modules
Functional enrichment analysis Functional enrichment analysis (TANGO)(TANGO) 714 enriched for GO ldquobiological 714 enriched for GO ldquobiological
processrdquoprocessrdquo 693 enriched for GO693 enriched for GO ldquocellular ldquocellular
compartmentrdquocompartmentrdquo 463 of known complexes enriched in 463 of known complexes enriched in
at least one BPMat least one BPM
Pivots tend to be Pivots tend to be multifunctionalmultifunctional
No of No of proteinsproteins
PivotsPivots ExpectedExpected SignificanceSignificance
GO complexesGO complexes 206206 2121 435435 pp=127=1271010-9-9
KEGG pathwaysKEGG pathways 7171 66 150150 pp=368=368 1010-3-3
Curated multi-Curated multi-complexed proteins complexed proteins (Krause et al 2006)(Krause et al 2006) 3939 88 055055 pp=749=749 1010-9-9
Pivot proteins are essentialPivot proteins are essential
Q Are proteins active in multiple partially Q Are proteins active in multiple partially redundant pathways more essentialredundant pathways more essential72124 (58) pivots essential (72124 (58) pivots essential (pp=142middot10=142middot10-23-23))Enrichment is not explained by node degrees Enrichment is not explained by node degrees ((pplt10lt10-5-5))Essential pivots closer in function to their BPMs Essential pivots closer in function to their BPMs than nonessential pivotsthan nonessential pivotsPivots significantly retained during evolution Pivots significantly retained during evolution (p=979middot10(p=979middot10-9-9))
SWR1 Ino80
SWR1Ino80 ExampleSWR1Ino80 ExamplePivots
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
BPM HuntingBPM Hunting
Look for pairs of pathways with GI Look for pairs of pathways with GI evidence for bufferingevidence for buffering
A pathway ndash a A pathway ndash a connectedconnected subnetwork in subnetwork in of PIsof PIs
Model scoring based on the density of GIsModel scoring based on the density of GIs
Log-likelihood scores accounting for GI Log-likelihood scores accounting for GI degreesdegrees
1 2
1 2
( )1 2 BPM1 2
1 2 null ( )
( ) (1 )(1 ( ))( | )( ) log log
( | ) ( ) (1 )(1 ( ))a b V V
genetica b a ba b V V
I a b I a bP V V MS V V
P V V M r I a b r I a b
AlgorithmicsAlgorithmics
High-scoring seedsHigh-scoring seeds Finding heavy bicliques Finding heavy bicliques
with connectivity with connectivity constraintsconstraints
Seed optimizationSeed optimization Greedy search Greedy search
maintaining maintaining connectivityconnectivity
Significance filteringSignificance filtering
Adding ldquopivot proteinsrdquo to Adding ldquopivot proteinsrdquo to BPMsBPMs
Pathway bifurcation Redundant sub-complexes
Pivot proteinsPivot proteins
Physically connected to both pathways Physically connected to both pathways in the BPMin the BPM
Connection significant given the Connection significant given the general network degreesgeneral network degrees
Analysis flowAnalysis flow
13632 genetic interactions
68172 physical interactions
140 models
124 pivot proteins in 40 models
BPM analysis
Pivot extraction
Physiological characteristics
bullEssentialitybullPhenotypesbullProtein AbundancebullCodon adaptation indexbullmRNA half-lifebullPhoshorylation
BPMs as functional modulesBPMs as functional modules
Functional enrichment analysis Functional enrichment analysis (TANGO)(TANGO) 714 enriched for GO ldquobiological 714 enriched for GO ldquobiological
processrdquoprocessrdquo 693 enriched for GO693 enriched for GO ldquocellular ldquocellular
compartmentrdquocompartmentrdquo 463 of known complexes enriched in 463 of known complexes enriched in
at least one BPMat least one BPM
Pivots tend to be Pivots tend to be multifunctionalmultifunctional
No of No of proteinsproteins
PivotsPivots ExpectedExpected SignificanceSignificance
GO complexesGO complexes 206206 2121 435435 pp=127=1271010-9-9
KEGG pathwaysKEGG pathways 7171 66 150150 pp=368=368 1010-3-3
Curated multi-Curated multi-complexed proteins complexed proteins (Krause et al 2006)(Krause et al 2006) 3939 88 055055 pp=749=749 1010-9-9
Pivot proteins are essentialPivot proteins are essential
Q Are proteins active in multiple partially Q Are proteins active in multiple partially redundant pathways more essentialredundant pathways more essential72124 (58) pivots essential (72124 (58) pivots essential (pp=142middot10=142middot10-23-23))Enrichment is not explained by node degrees Enrichment is not explained by node degrees ((pplt10lt10-5-5))Essential pivots closer in function to their BPMs Essential pivots closer in function to their BPMs than nonessential pivotsthan nonessential pivotsPivots significantly retained during evolution Pivots significantly retained during evolution (p=979middot10(p=979middot10-9-9))
SWR1 Ino80
SWR1Ino80 ExampleSWR1Ino80 ExamplePivots
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
AlgorithmicsAlgorithmics
High-scoring seedsHigh-scoring seeds Finding heavy bicliques Finding heavy bicliques
with connectivity with connectivity constraintsconstraints
Seed optimizationSeed optimization Greedy search Greedy search
maintaining maintaining connectivityconnectivity
Significance filteringSignificance filtering
Adding ldquopivot proteinsrdquo to Adding ldquopivot proteinsrdquo to BPMsBPMs
Pathway bifurcation Redundant sub-complexes
Pivot proteinsPivot proteins
Physically connected to both pathways Physically connected to both pathways in the BPMin the BPM
Connection significant given the Connection significant given the general network degreesgeneral network degrees
Analysis flowAnalysis flow
13632 genetic interactions
68172 physical interactions
140 models
124 pivot proteins in 40 models
BPM analysis
Pivot extraction
Physiological characteristics
bullEssentialitybullPhenotypesbullProtein AbundancebullCodon adaptation indexbullmRNA half-lifebullPhoshorylation
BPMs as functional modulesBPMs as functional modules
Functional enrichment analysis Functional enrichment analysis (TANGO)(TANGO) 714 enriched for GO ldquobiological 714 enriched for GO ldquobiological
processrdquoprocessrdquo 693 enriched for GO693 enriched for GO ldquocellular ldquocellular
compartmentrdquocompartmentrdquo 463 of known complexes enriched in 463 of known complexes enriched in
at least one BPMat least one BPM
Pivots tend to be Pivots tend to be multifunctionalmultifunctional
No of No of proteinsproteins
PivotsPivots ExpectedExpected SignificanceSignificance
GO complexesGO complexes 206206 2121 435435 pp=127=1271010-9-9
KEGG pathwaysKEGG pathways 7171 66 150150 pp=368=368 1010-3-3
Curated multi-Curated multi-complexed proteins complexed proteins (Krause et al 2006)(Krause et al 2006) 3939 88 055055 pp=749=749 1010-9-9
Pivot proteins are essentialPivot proteins are essential
Q Are proteins active in multiple partially Q Are proteins active in multiple partially redundant pathways more essentialredundant pathways more essential72124 (58) pivots essential (72124 (58) pivots essential (pp=142middot10=142middot10-23-23))Enrichment is not explained by node degrees Enrichment is not explained by node degrees ((pplt10lt10-5-5))Essential pivots closer in function to their BPMs Essential pivots closer in function to their BPMs than nonessential pivotsthan nonessential pivotsPivots significantly retained during evolution Pivots significantly retained during evolution (p=979middot10(p=979middot10-9-9))
SWR1 Ino80
SWR1Ino80 ExampleSWR1Ino80 ExamplePivots
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
Adding ldquopivot proteinsrdquo to Adding ldquopivot proteinsrdquo to BPMsBPMs
Pathway bifurcation Redundant sub-complexes
Pivot proteinsPivot proteins
Physically connected to both pathways Physically connected to both pathways in the BPMin the BPM
Connection significant given the Connection significant given the general network degreesgeneral network degrees
Analysis flowAnalysis flow
13632 genetic interactions
68172 physical interactions
140 models
124 pivot proteins in 40 models
BPM analysis
Pivot extraction
Physiological characteristics
bullEssentialitybullPhenotypesbullProtein AbundancebullCodon adaptation indexbullmRNA half-lifebullPhoshorylation
BPMs as functional modulesBPMs as functional modules
Functional enrichment analysis Functional enrichment analysis (TANGO)(TANGO) 714 enriched for GO ldquobiological 714 enriched for GO ldquobiological
processrdquoprocessrdquo 693 enriched for GO693 enriched for GO ldquocellular ldquocellular
compartmentrdquocompartmentrdquo 463 of known complexes enriched in 463 of known complexes enriched in
at least one BPMat least one BPM
Pivots tend to be Pivots tend to be multifunctionalmultifunctional
No of No of proteinsproteins
PivotsPivots ExpectedExpected SignificanceSignificance
GO complexesGO complexes 206206 2121 435435 pp=127=1271010-9-9
KEGG pathwaysKEGG pathways 7171 66 150150 pp=368=368 1010-3-3
Curated multi-Curated multi-complexed proteins complexed proteins (Krause et al 2006)(Krause et al 2006) 3939 88 055055 pp=749=749 1010-9-9
Pivot proteins are essentialPivot proteins are essential
Q Are proteins active in multiple partially Q Are proteins active in multiple partially redundant pathways more essentialredundant pathways more essential72124 (58) pivots essential (72124 (58) pivots essential (pp=142middot10=142middot10-23-23))Enrichment is not explained by node degrees Enrichment is not explained by node degrees ((pplt10lt10-5-5))Essential pivots closer in function to their BPMs Essential pivots closer in function to their BPMs than nonessential pivotsthan nonessential pivotsPivots significantly retained during evolution Pivots significantly retained during evolution (p=979middot10(p=979middot10-9-9))
SWR1 Ino80
SWR1Ino80 ExampleSWR1Ino80 ExamplePivots
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
Pivot proteinsPivot proteins
Physically connected to both pathways Physically connected to both pathways in the BPMin the BPM
Connection significant given the Connection significant given the general network degreesgeneral network degrees
Analysis flowAnalysis flow
13632 genetic interactions
68172 physical interactions
140 models
124 pivot proteins in 40 models
BPM analysis
Pivot extraction
Physiological characteristics
bullEssentialitybullPhenotypesbullProtein AbundancebullCodon adaptation indexbullmRNA half-lifebullPhoshorylation
BPMs as functional modulesBPMs as functional modules
Functional enrichment analysis Functional enrichment analysis (TANGO)(TANGO) 714 enriched for GO ldquobiological 714 enriched for GO ldquobiological
processrdquoprocessrdquo 693 enriched for GO693 enriched for GO ldquocellular ldquocellular
compartmentrdquocompartmentrdquo 463 of known complexes enriched in 463 of known complexes enriched in
at least one BPMat least one BPM
Pivots tend to be Pivots tend to be multifunctionalmultifunctional
No of No of proteinsproteins
PivotsPivots ExpectedExpected SignificanceSignificance
GO complexesGO complexes 206206 2121 435435 pp=127=1271010-9-9
KEGG pathwaysKEGG pathways 7171 66 150150 pp=368=368 1010-3-3
Curated multi-Curated multi-complexed proteins complexed proteins (Krause et al 2006)(Krause et al 2006) 3939 88 055055 pp=749=749 1010-9-9
Pivot proteins are essentialPivot proteins are essential
Q Are proteins active in multiple partially Q Are proteins active in multiple partially redundant pathways more essentialredundant pathways more essential72124 (58) pivots essential (72124 (58) pivots essential (pp=142middot10=142middot10-23-23))Enrichment is not explained by node degrees Enrichment is not explained by node degrees ((pplt10lt10-5-5))Essential pivots closer in function to their BPMs Essential pivots closer in function to their BPMs than nonessential pivotsthan nonessential pivotsPivots significantly retained during evolution Pivots significantly retained during evolution (p=979middot10(p=979middot10-9-9))
SWR1 Ino80
SWR1Ino80 ExampleSWR1Ino80 ExamplePivots
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
Analysis flowAnalysis flow
13632 genetic interactions
68172 physical interactions
140 models
124 pivot proteins in 40 models
BPM analysis
Pivot extraction
Physiological characteristics
bullEssentialitybullPhenotypesbullProtein AbundancebullCodon adaptation indexbullmRNA half-lifebullPhoshorylation
BPMs as functional modulesBPMs as functional modules
Functional enrichment analysis Functional enrichment analysis (TANGO)(TANGO) 714 enriched for GO ldquobiological 714 enriched for GO ldquobiological
processrdquoprocessrdquo 693 enriched for GO693 enriched for GO ldquocellular ldquocellular
compartmentrdquocompartmentrdquo 463 of known complexes enriched in 463 of known complexes enriched in
at least one BPMat least one BPM
Pivots tend to be Pivots tend to be multifunctionalmultifunctional
No of No of proteinsproteins
PivotsPivots ExpectedExpected SignificanceSignificance
GO complexesGO complexes 206206 2121 435435 pp=127=1271010-9-9
KEGG pathwaysKEGG pathways 7171 66 150150 pp=368=368 1010-3-3
Curated multi-Curated multi-complexed proteins complexed proteins (Krause et al 2006)(Krause et al 2006) 3939 88 055055 pp=749=749 1010-9-9
Pivot proteins are essentialPivot proteins are essential
Q Are proteins active in multiple partially Q Are proteins active in multiple partially redundant pathways more essentialredundant pathways more essential72124 (58) pivots essential (72124 (58) pivots essential (pp=142middot10=142middot10-23-23))Enrichment is not explained by node degrees Enrichment is not explained by node degrees ((pplt10lt10-5-5))Essential pivots closer in function to their BPMs Essential pivots closer in function to their BPMs than nonessential pivotsthan nonessential pivotsPivots significantly retained during evolution Pivots significantly retained during evolution (p=979middot10(p=979middot10-9-9))
SWR1 Ino80
SWR1Ino80 ExampleSWR1Ino80 ExamplePivots
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
BPMs as functional modulesBPMs as functional modules
Functional enrichment analysis Functional enrichment analysis (TANGO)(TANGO) 714 enriched for GO ldquobiological 714 enriched for GO ldquobiological
processrdquoprocessrdquo 693 enriched for GO693 enriched for GO ldquocellular ldquocellular
compartmentrdquocompartmentrdquo 463 of known complexes enriched in 463 of known complexes enriched in
at least one BPMat least one BPM
Pivots tend to be Pivots tend to be multifunctionalmultifunctional
No of No of proteinsproteins
PivotsPivots ExpectedExpected SignificanceSignificance
GO complexesGO complexes 206206 2121 435435 pp=127=1271010-9-9
KEGG pathwaysKEGG pathways 7171 66 150150 pp=368=368 1010-3-3
Curated multi-Curated multi-complexed proteins complexed proteins (Krause et al 2006)(Krause et al 2006) 3939 88 055055 pp=749=749 1010-9-9
Pivot proteins are essentialPivot proteins are essential
Q Are proteins active in multiple partially Q Are proteins active in multiple partially redundant pathways more essentialredundant pathways more essential72124 (58) pivots essential (72124 (58) pivots essential (pp=142middot10=142middot10-23-23))Enrichment is not explained by node degrees Enrichment is not explained by node degrees ((pplt10lt10-5-5))Essential pivots closer in function to their BPMs Essential pivots closer in function to their BPMs than nonessential pivotsthan nonessential pivotsPivots significantly retained during evolution Pivots significantly retained during evolution (p=979middot10(p=979middot10-9-9))
SWR1 Ino80
SWR1Ino80 ExampleSWR1Ino80 ExamplePivots
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
Pivots tend to be Pivots tend to be multifunctionalmultifunctional
No of No of proteinsproteins
PivotsPivots ExpectedExpected SignificanceSignificance
GO complexesGO complexes 206206 2121 435435 pp=127=1271010-9-9
KEGG pathwaysKEGG pathways 7171 66 150150 pp=368=368 1010-3-3
Curated multi-Curated multi-complexed proteins complexed proteins (Krause et al 2006)(Krause et al 2006) 3939 88 055055 pp=749=749 1010-9-9
Pivot proteins are essentialPivot proteins are essential
Q Are proteins active in multiple partially Q Are proteins active in multiple partially redundant pathways more essentialredundant pathways more essential72124 (58) pivots essential (72124 (58) pivots essential (pp=142middot10=142middot10-23-23))Enrichment is not explained by node degrees Enrichment is not explained by node degrees ((pplt10lt10-5-5))Essential pivots closer in function to their BPMs Essential pivots closer in function to their BPMs than nonessential pivotsthan nonessential pivotsPivots significantly retained during evolution Pivots significantly retained during evolution (p=979middot10(p=979middot10-9-9))
SWR1 Ino80
SWR1Ino80 ExampleSWR1Ino80 ExamplePivots
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
Pivot proteins are essentialPivot proteins are essential
Q Are proteins active in multiple partially Q Are proteins active in multiple partially redundant pathways more essentialredundant pathways more essential72124 (58) pivots essential (72124 (58) pivots essential (pp=142middot10=142middot10-23-23))Enrichment is not explained by node degrees Enrichment is not explained by node degrees ((pplt10lt10-5-5))Essential pivots closer in function to their BPMs Essential pivots closer in function to their BPMs than nonessential pivotsthan nonessential pivotsPivots significantly retained during evolution Pivots significantly retained during evolution (p=979middot10(p=979middot10-9-9))
SWR1 Ino80
SWR1Ino80 ExampleSWR1Ino80 ExamplePivots
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
SWR1 Ino80
SWR1Ino80 ExampleSWR1Ino80 ExamplePivots
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
SAGANuclear pore exampleSAGANuclear pore example
Nuclear pore SAGA
Pivots
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
Physiological properties of Physiological properties of BPMsBPMs
Proteins studiedProteins studied 850 proteins within BPMs850 proteins within BPMs 120 pivot proteins 120 pivot proteins Focus onFocus on mRNA half lifemRNA half life of phosphorylation sites of phosphorylation sites
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
BPMs are strongly regulatedBPMs are strongly regulatedGenes within BPMsGenes within BPMs ShorterShorter mRNA half-life ( mRNA half-life (pp=19=191010-9-9)) MoreMore phosphorylation sites ( phosphorylation sites (pp=63=631010-9-9))
Both properties may represent regulationBoth properties may represent regulation
Redundant pathways Redundant pathways Strict regulation Strict regulation
Enrichments not explained by degrees Enrichments not explained by degrees essentiality any enriched functionessentiality any enriched function
mRNA half-life experimental data mRNA half-life experimental data phosphosites predictedphosphosites predicted
Properties generally not correlatedProperties generally not correlated
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
Next generation ndash quantitative GIsNext generation ndash quantitative GIs
Data on Data on quantitativequantitative genetic interactions genetic interactions scores is becoming availablescores is becoming available
These include also These include also alleviatingalleviating interactions interactions
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
Alleviating interactionsAlleviating interactions
The observed phenotype is The observed phenotype is betterbetter than than what we would ldquoexpectrdquowhat we would ldquoexpectrdquoldquoldquoExpectrdquo ndash multiplicative modelExpectrdquo ndash multiplicative model Mutation A ndash 80 fitness is retainedMutation A ndash 80 fitness is retained Mutation B ndash 60 fitness is retainedMutation B ndash 60 fitness is retained A amp B ndash 48 fitness is retainedA amp B ndash 48 fitness is retained
The first deletion ruins the functionally of The first deletion ruins the functionally of the pathway and the second one does not the pathway and the second one does not have an effecthave an effectAn alleviating interaction is a sign of An alleviating interaction is a sign of participation in the same pathwayparticipation in the same pathway
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Probabilistic model for the Probabilistic model for the aggravatingalleviating interaction scoresaggravatingalleviating interaction scores
Detection of pathways and buffering using Detection of pathways and buffering using quantitative dataquantitative data
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
Next challenge simultaneously extraction Next challenge simultaneously extraction of multiple pathwaysof multiple pathways
The output includesThe output includes Pathway boundariesPathway boundaries Pairs of buffering pathwaysPairs of buffering pathways
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
Extracting BPMs from quantitative Extracting BPMs from quantitative datadata
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis
SummarySummaryComputational analysis outlines pathways and Computational analysis outlines pathways and buffering pathway pairs - BPMsbuffering pathway pairs - BPMs
BPM pathways tend to be strictly regulatedBPM pathways tend to be strictly regulated
BPM pivots correspond to essential BPM pivots correspond to essential multifunctional proteinsmultifunctional proteins
Quantitative GIs carry the promise of a Quantitative GIs carry the promise of a reacher analysisreacher analysis