Scalable metabolic reconstructionfor metagenomic data

and the human microbiome

Sahar Abubucker, Nicola Segata, Johannes Goll,Alyxandria Schubert, Beltran Rodriguez-Mueller, Jeremy Zucker,

the Human Microbiome Project Metabolic Reconstruction team,the Human Microbiome Consortium,

Patrick D. Schloss, Dirk Gevers, Makedonka Mitreva,Curtis Huttenhower

04-14-11Harvard School of Public HealthDepartment of Biostatistics

2

What’s metagenomics?Total collection of microorganisms

within a community

Also microbial community or microbiota

Total genomic potential of a microbial community

Total biomolecular repertoire of a microbial community

Study of uncultured microorganisms from the environment, which can include

humans or other living hosts

The Human Microbiome Project for a normal population

300 People/15(18) Body

Sites

Multifaceted analyses

2 clin. centers, 4 seq. centers, data generation,technology development, computational tools,

ethics…

Multifaceted data

Human population

Microbial population

Novel organisms

BiotypesVirusesMetabolism

>12,000 samples>50M 16S seqs.4.6Tbp unique

metagenomicsequence

>1,900 referencegenomes

Full clinicalmetadata

15+ Demonstration Projects for microbial communities in disease

Gastrointestinal

All include additional subjects and technology development

Skin

PsoriasisAcneAtopic

dermatitis

ObesityCrohn’s diseaseUlcerative

colitisAutoimmunityCancerNecrotizing

enterocolitis

Urogenital

Bacterialvaginosis

STDsReproductive

health

5

What to do with your metagenome?

(x1010)

Diagnostic or prognostic

biomarker for host disease

Public health tool monitoring

population health and interactions

Comprehensive snapshot of

microbial ecology and evolution

Reservoir of gene and protein

functional informationWho’s there?

What are they doing?

What do functional genomic data tell us about microbiomes?

What can our microbiomes tell us about us?

6

GeneexpressionSNPgenotypes

Metabolic/Functional Reconstruction: The Goal

Healthy/IBDBMI

Diet

Taxon abundancesEnzyme family abundancesPathway abundances

Batch effects?Populationstructure?

Niches &Phylogeny

Test for correlates

Multiplehypothesiscorrection

Featureselection

p >> n

Confounds/stratification/environment

Cross-validate

Biological story?

Independent sample

Intervention/perturbation

7

HMP: Metabolic reconstruction

WGS reads

Pathways/modules

Genes(KOs)

Pathways(KEGGs)

Functional seq.KEGG + MetaCYC

CAZy, TCDB,VFDB, MEROPS…

BLAST → Genes

rra

r

raa

p

gap

ggc

)(

)(

1

)()1(

||

1)(

Genes → PathwaysMinPath (Ye 2009)

SmoothingWitten-Bell

otherwiseTNNgc

gcTNTVTNgc

)/()(

0)()/()/()(Gap filling

c(g) = max( c(g), median )

300 subjects1-3 visits/subject~6 body sites/visit

10-200M reads/sample100bp reads

BLAST

?Taxonomic limitation

Rem. paths in taxa < ave.

XipeDistinguish zero/low

(Rodriguez-Mueller in review)

HUMAnN:HMP UnifiedMetabolic AnalysisNetworkhttp://huttenhower.sph.harvard.edu/humann

8

HMP: Metabolic reconstruction

Pathway coverage Pathway abundance

9

HUMAnN: Validating gene and pathwayabundances on synthetic data

Validated on individual gene families, module coverage, and abundance

• 4 synthetic communities:Low (20 org.) and high (100 org.)

complexityEven and lognormal abundances

• Few false negatives: short genes (<100bp),

taxonomically rare pathways

• Few false positives: large and multicopy

(not many in bacteria)

Individual gene families

ρ=0.86

10

Functional modules in 741 HMP samples

Coverage

Abundance

ANO(BM)PF O(SP)S RCO(TD)

← Samples →

← P

ath

wa

ys

→

• Zero microbes (of ~1,000)

are core among body sites

• Zero microbes are core

among individuals

• 19 (of ~220) pathways are

present in every sample

• 53 pathways are present in

90%+ samples

• Only 31 (of 1,110) pathways

are present/absent from

exactly one body site

• 263 pathways aredifferentially

abundant inexactly one body

site

11

A portrait of the human microbiome:Who’s there?

With Jacques Izard, Susan Haake, Katherine Lemon

12

Pat

hway

cov

erag

e

A portrait of the human microbiome:What are they doing?

13

HMP: How do microbes vary within each body site across the population?

14

HMP: How do body sites compare between individuals across the population?

15

HMP: Penetrance of species (OTUs)across the population

Data from Pat Schloss

16

HMP: Penetrance of genera (phylotypes)across the population

Data from Pat Schloss

17

HMP: Penetrance of pathwaysacross the population

KEGG Metabolic modulesM00001: Glycolysis (Embden-Meyerhof)

M00002: Glycolysis, core moduleM00003: Gluconeogenesis

M00004: Pentose phosphate cycleM00007: Pentose phosphate (non-oxidative)

M00049: Adenine biosynthesisM00050: Guanine biosynthesis

M00052: Pyrimidine biosynthesisM00053: Pyrimidine deoxyribonuleotide biosynthesis

M00120: Coenzyme A biosynthesisM00126: Tetrahydrofolate biosynthesis

M00157: F-type ATPase, bacteriaM00164: ATP synthase

M00178: Ribosome, bacteriaM00183: RNA polymerase, bacteria

M00260: DNA polymerase III complex, bacteriaM00335: Sec (secretion) system

M00359: Aminoacyl-tRNA biosynthesisM00360: Aminoacyl-tRNA biosynthesis, prokaryotes

M00362: Nucleotide sugar biosynthesis, prokaryotesM00006: Pentose phosphate (oxidative)

M00051: Uridine monophosphate biosynthesisM00125: Riboflavin biosynthesis

M00008: Entner-Doudoroff pathwayM00239: Peptides/nickel transport system

M00018: Threonine biosynthesisM00168: CAM (Crassulacean acid metabolism), dark

M00167: Reductive pentose phosphate cycle

• Human microbiome functional structure dictated

primarily by microbial niche, not host (in health)

• Huge variation among hosts in who’s there;

small variation in what they’re doing

• Note: definitely variation in how thesefunctions are implemented

• Does not yet speak in detail to hostenvironment (diet!), genetics, or disease

18

Population summary statistics population biology

Posterior fornix, ref. genomes← Individuals →

← S

pe

cie

s →

← P

ath

wa

ys

→

Lactobacillus inersLactobacillus crispatusGardnerella vaginalisLactobacillus jenseniiLactobacillus gasseri

Posterior fornix, functional modules

Basic biology, sugar transport

Essential amino acids

Urea cycle, amines, aromatic AAs

19

LEfSe: Metagenomic classcomparison and explanation

LEfSe

http://huttenhower.sph.harvard.edu/lefse

Nicola Segata

LDA +Effect Size

20

LEfSe: Evaluation on synthetic data

Their FP rate

Our FP rate

21

Microbes characteristic of theoral and gut microbiota

Aerobic, microaerobic and anaerobic communities

• High oxygen:skin, nasal

• Mid oxygen:vaginal, oral

• Low oxygen:gut

23

LEfSe: The TRUC murine colitis microbiotaWith Wendy Garrett

24

Microbial biomolecular function and biomarkers in the human microbiome: the story so far?

• Who’s there changes– What they’re doing doesn’t (as much)– How they’re doing it does

• The data so far only scratch the surface– Only 1/3 to 2/3 of the reads/sample map to cataloged gene families– Only 1/3 to 2/3 of these gene families have cataloged functions– Very much in line with MetaHIT study of gut microbiota– Job security!

• Looking forward to functional reconstruction…– In environmental communities– With respect to host environment + genetics– With respect to host disease

25

Thanks!

Jacques IzardWendy Garrett

Pinaki SarderNicola Segata

Levi Waldron LarisaMiropolsky

Interested? We’re recruiting graduate students and postdocs!

Human Microbiome Project

HMP Metabolic Reconstruction

George WeinstockJennifer WortmanOwen WhiteMakedonka MitrevaErica SodergrenMihai PopVivien Bonazzi Jane PetersonLita Proctor

Sahar AbubuckerYuzhen Ye

Beltran Rodriguez-MuellerJeremy ZuckerQiandong Zeng

Mathangi ThiagarajanBrandi Cantarel

Maria RiveraBarbara Methe

Bill KlimkeDaniel Haft

Ramnik Xavier Dirk Gevers

Bruce Birren Mark DalyDoyle Ward Eric AlmAshlee Earl Lisa Cosimi

http://huttenhower.sph.harvard.eduhttp://huttenhower.sph.harvard.edu/humann

http://huttenhower.sph.harvard.edu/lefse