49
Overview of single cell RNA-Seq analysis Vishal Koparde, Ph. D. CCR Collaborative Bioinformatics Resource (CCBR) Leidos Biomedical Research, Inc.

Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

  • Upload
    others

  • View
    3

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

OverviewofsinglecellRNA-Seq analysis

VishalKoparde,Ph.D.CCRCollaborativeBioinformaticsResource(CCBR)

Leidos BiomedicalResearch,Inc.

Page 2: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Outline

• BulkRNA-Seq vsscRNA-Seq• scRNA-Seq applications&challenges• scRNA-Seq assays• scRNA-Seq pipeline(UMI,QC,downstreamanalysis)• t-SNEvisualization• CCBRscRNA-Seq pipeline

Page 3: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

BulkRNA-Seq:Averagingmaskscellularheterogeneity

Page 4: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

BulkRNA-Seq

• Thebasicunitofresearchisthecell.Butweareusuallyanalyzingpopulationsofcellsandthiscan:• Leadtofalsepositivesfromunderestimatingbiologicalvariability• Missimportantbiologicaldivisions

Mean=.959Stddev=.048SampleSize=2

Page 5: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

BulkRNA-Seq

• Impossibletoconcludeifthesedifferencesareduetoauniformpropertyofallcellswithinatumor/genegrouporduetoadditiveeffectsofjustafewsub-cloneswithinatumor/genegroup

• ConclusionsfrombulkanalysiscanberepresentativeofNOTHING

Page 6: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

scRNA-Seq

Page 7: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

scRNA-Seq evolution

scRNA-Seq hasbeenaroundforadecade,buthasnowmaturedwithpracticalapplications

Page 8: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

PotentialofscRNA-Seq

• InvestigateTranscriptHeterogeneity• Differencesintranscriptabundance• Usageofalternatetranscriptioninitiationandpolyadenylationsites• Alternativesplicinganddifferentialexpressionoftranscriptisoforms

Page 9: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Challenges:scRNA-Seq

• Limitedamountsofmaterial• Illumina/PacBio require1ng-500ng• 1bacteriumor1cancercellcontain1fg-1pg• ~10-50pgoftotalRNAà 1-5%ofwhichismRNA• WTA/WGAbecomereallyimportant

Page 10: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Challenges:scRNA-Seq

• Relativelyhighcorrelationisexpectedbetweensamples

• Highlyvariableandnoisy• Manydifferencesevenbetweensametypeofcells• Biologicalsignaldiminishesatcellularlevelandbecomescomparabletotechnicalvariability

Page 11: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Challenges:scRNA-Seq

• Viability• Cellstress• Dependsonapplication• Budgetconsiderations• Technicalvariation• Cellcapture• Librarypreparation• Inherentsequencingbiases

Page 12: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

scRNA-Seq:Differentassays

• TherearemanyscRNA-Seqassaysoutthere:• Somearecommercialized… growing• Fulltranscriptvs3’• Isoformvsgene• Lessormoreautomated• Differentlevelsofthroughput• Differencesincosts• Differencesindownstreambioinformaticsanalysis

Page 13: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

SmartSeq2

• FulltranscriptscRNA-Seq• Selectsforpoly-Atail• 5’switchingforfulltranscriptlengthcapture• StandardIlluminasequencing• Off-the-shelfproducts

• Hundredsofsamples• NoUMIsused

Page 14: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

DropSeq

• Droplet-based• Throughputfromhundredstothousands• 3’Endonly• Singlecelltranscriptomesattachedtomicroparticles• UMIs(uniquemolecularidentifiers)• Cellbarcodes• Drop-seq toolssoftware

Page 15: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

10XGenomics

• Droplet-based• GEMà GelbeadinEMulsion

• 3’mRNA• Standardizedinstrumentationandreagents• Morehigh-throughputscalingtotensofthousandsofcells• Lessprocessingtime• Cellrangersoftware

Page 16: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

10XGenomicsCellRanger

• cellranger mkfastq … BCLtofastq• cellranger count… fastqtocountmatrix• cellranger aggr … combinefastqs frommultiplerunsintoonecountmatrix• cellranger reanalyze… countmatrixtodimensionalityreduction/clustering• PythonandRsupportincluded

Page 17: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Analysispipelinesdifferalot… why?

• scRNA-Seqdatasetscomeindifferentflavors:• “Fishingexpedition”:Dissociatetissueà getsubpopulations!• CasevsControl• Moleculartransformationsovertime(Pseudotime)• Perturbationseg.CRISPRScreening

• Veryhardtohaveasingledo-it-allpipelineforscRNA-Seqanalysis

Page 18: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Newpipelineskeeppoppingup…

https://github.com/seandavi/awesome-single-cell

Page 19: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Rawdata

Page 20: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

AssaysdifferbyFASTQcontent

• Dropseq and10XisreallyjustsingleendRNASeqonapercelllevel

Page 21: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

UMI

• UniqueMolecularIdentifiersareshort(4-10bp)randombarcodesaddedtotranscriptsduringreverse-transcription.• TheyenablesequencingreadstobeassignedtoindividualtranscriptmoleculesandthustheremovalofamplificationnoiseandbiasesfromscRNASeqdata

Page 22: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

IssueswithUMI

• DifferentUMIdoesn’tnecessarilymeandifferentmolecule

• Differenttranscriptdoesn’tnecessarilymeandifferentmolecule

• SameUMIdoesn’tnecessarilymeansamemolecule

• Cell-rangeperformsbuilt-inerrorcorrectionforUMI-Read-Transcriptmapping

Page 23: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

scRNA-Seq pipeline:QC

• ReadalignmentsandmuchoftheQCcanbehandledsimilartobulkRNA-Seq

• Cellrangerusesmaskedhumanandmousegenometoexclude“problem”regionsapriori

Page 24: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

scRNA-Seq pipeline:QC

Page 25: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

scRNA-Seq pipeline:Downstreamanalysis

• Completeend-to-endpipelineinR• SINCERAusedbyCincinnatiChildren’sHospital

• Rpackages• Seurat• scater• SC3

Page 26: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

CommonstepsofascRNA-Seq pipeline

• Aligningtogenome• QC• Readcounting• Filtering• Normalization• Clustering

• Selectionoftoolsvarybetweendatasettodataset… dependonthebiggerbiologicalquestion

Page 27: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

CountMatrix

• 5000-12000genespercellon10X

• MostentriesarezeroàSparsematrix

• Differentefficientrepresentationofsparsematrixinmemoryareused

Page 28: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

CountMatrixascoordinatelist

Page 29: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Geneshavedifferentdistributions

Page 30: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Geneshavedifferentdistributions

Page 31: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Filtering:Numberofreads

• Cellbarcodeswithlownumberofreadsareexcluded

user-defined

arbitrarycutoff

Page 32: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Filtering:Numberofgenes

user-defined

arbitrarycutoff

Page 33: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Filtering:NumberofUMIs

user-defined

arbitrarycutoff

Page 34: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Filtering:Mitochondrial%

user-defined

arbitrarycutoff

Page 35: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Filtering:Outliersare“multipleoffenders”

user-defined

arbitrarycutoff

Page 36: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

AutomaticFiltering

Page 37: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Normalization

• scRNASeqassaysarepronetoconfoundingbatcheffects,hencewenormalize• ManyRNA-Seqnormalizationtechniquescandirectlybeapplied• Exploringtheideaofcombiningorselectingfromacollectionofnormalizationorcorrectionmethodsbestforaspecificcase• SomebelievethatUMIbaseanalysisneednotbenormalizedbetweensamplesasabsolutecountofmoleculesarereported

Page 38: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Normalization

• No.ofreadspercellvaryhighlyandlibrarysizenormalizationisrequired• MethodslikeRSEMalreadyincorporatelibrarysizeanddonotrequirenormalization• “normalizationfactor”isamultiplicationfactor,whichisanestimateofthelibrarysizerelativetotheothercells• TMM,UQ,RLE

Page 39: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

NormalizationRaw

CPM RLEUQ

Page 40: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Visualization

• PCA(lineardimensionalityreduction)• Smallpercentageofvarianceisexplainedbyfirst2or3PCs

• t-SNE(non-lineardimensionalityreduction)coloredwithk-meansclustering• Notstraightforwardtodeterminek• kcanbesettothenumberofsignificantPCswhichcanbedeterminedbyajackstrawplot(computationallyintensive)• Iterativeprocesswithgradualfinetuning

Page 41: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

PCAvs.t-SNE

Page 42: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Whatist-SNE?

• t-DistributedStochasticNeighborEmbedding• “Dimensionalityreduction”-->representmulti-dimensionalitydatain2or3-Dspace• PCA:• Linear• Unabletointerpretcomplex

polynomialrelationshipbetweenfeatures

• Focusesonrepresentingdissimilarpointsfartherawayinlowerdimensions

• t-SNE:• Non-linear• Focusesonrepresentingsimilar

datapoints closetogetherinlowerdimensions

• notcapableofretainingboththelocalandglobalstructureofthedataatthesametime

Page 43: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Thingstorememberaboutt-SNE

• Canarriveatdifferentsolutionsdependingontheinitial“random”distributions… Hence,itisimportanttousethesame“seed”everytime

• Notcapableofretainingboththelocalandglobalstructureofthedataatthesametime

• OneoftheunderlyingalgorithmsusedinthelatestiPhoneXFacialrecognitionsoftware

Algorithm FERaccuracy

PCA 75.40%

LDA 75.90%

LLE 87.70%

SNE 90.60%

t-SNE 94.50%

Page 44: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

CCBRscRNASeqpipeline

Development:https://github.com/CCBR/scRNASeq Production:https://github.com/CCBR/Pipeliner

Page 45: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

CCBRscRNASeqpipeline

• Startingpoints• 10Xgenomicsfastq• 10Xgenomicscountmatrix

• DataFilteringetc.• Downstreamanalysis

• k-meansclustering• PCA• tSNE plot• markergenelists

• Alltasksaresubmittedasslurm jobstobiowulf

Page 46: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

CCBRPipeliner:examplevisualizations

Page 47: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

CCBRPipeliner:What’sinthe“pipeline”?

• Futureenhancements:• ComprehensiveQCatacellularlevel

• Smoothmergingofmultipledatasetswithminimalbatcheffect(newerversionofSeurat)

• Pseudotimeanalysis

• Shinyappforinteractive3-Dt-SNEforeffectivevisualizationwhichwillallowpseudocoloringbasedon:• Clusternumber• Gene• Genesets /Pathways

Page 48: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

Takehomemessage

• “scRNA Seqanditsanalysis”fieldischangingatarapidpace• ManyofthethingsthatyouheardaboutscRNA Seqanalysistodaywillbeobsoleteinafewmonths!

THANKYOUFORYOURATTENTION

Page 49: Overview of single cell RNA-Seqanalysis · Bulk RNA-Seq •The basic unit of research is the cell. But we are usually analyzing populations of cells and this can: •Lead to false

ContactUs

• E-mail• [email protected]

• Website• http://bioinformatics.cancer.gov

• Blg37/R3041