1. Charting the function of microbes and microbial communities Curtis Huttenhower 11-17-11 Harvard School of Public Health Department of Biostatistics

2. Valm et al, PNAS 2011

3. What to do with your metagenome? 3 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 information Who’s there? What are they doing? Who’s there varies: your microbiota is plastic and personalized. This personalization is true at the level of phyla, genera, species, strains, and sequence variants. What they’re doing is adapting to their environment: you, your body, and your environment.

4. The NIH Human Microbiome Project (HMP): A comprehensive microbial survey What is a “normal” human microbiome? 300 healthy human subjects Multiple body sites 15 male, 18 female Multiple visits Clinical metadata www.hmpdacc.org Slides by Dirk Gevers

5. A three-tier study design… 16S 16S WGSWGS ref ref

6. …for mining metagenomic data contigs pathways ~100M reads per sample Assembly Annotation Map on ~50% ~90M proteins 16S 16S WGSWGS Filtering/ trimming Chimera removal >3k reads per sample BLAST against functional DBs Organismal census at different taxonomic levels ref ref Taxonomic classification (RDP) Clustering into OTUs census... ~36% ~57% genes

7. “Pathogen” carriage varies a lot 7 Gardnerella Alistipes Capnocytophaga Actinomyces Gemella 22 ***uniquely identifiable*** nonzero abundance “pathogens” from NIAID’s list of 135 +Propionibacterium >0.66

8. Phenotypes that explain variation (or not) can be surprising 8 Normalized relative abundance

9. Phenotypes that explain variation (or not) can be surprising 9 Normalized relative abundance

10. Phenotypes that explain variation (or not) can be surprising 10 Normalized relative abundance

11. Gene expression SNP genotypes A functional perspective on the human microbiome 11 Healthy/IBD BMI Diet Taxon abundances Enzyme family abundances Pathway abundances Functional seq. KEGG + MetaCYC CAZy, TCDB, VFDB, MEROPS… 100 subjects 1-3 visits/subject ~7 body sites/visit 10-200M reads/sample 100bp reads Metagenomic reads Enzymes and pathways ? HUMAnN HMP Unified Metabolic Analysis Network http://huttenhower.sph.harvard.edu/humann BLAST

12. HUMAnN: Metabolic reconstruction 12 Pathway coveragePathway abundance ← Samples → ← Pathways→ VaginalSkinNaresGutOral (SupP)Oral (BM)Oral (TD) ← Pathways→ ← Samples → VaginalSkinNaresGutOral (SupP)Oral (BM)Oral (TD)

13. ← Subjects → ← Pathway abundance → ← Phylotype abundance → A portrait of the healthy human microbiome: Who’s there vs. what they’re doing 13 VaginalSkinNaresGut Oral (SupP)Oral (BM)Oral (TD) ← Phylotype abundance → ← Subjects → ← Pathway abundance →

14. ← ~700 HMP communities→ Niche specialization in human microbiome function 14 Metabolic modules in the KEGG functional catalog enriched at one or more body habitats 16 (of 251) modules strongly “core” at 90%+ coverage in 90%+ individuals at 7 body sites 24 modules at 33%+ coverage 71 modules (28%) weakly “core” at 33%+ coverage in 66%+ individuals at 6+ body sites Contrast zero phylotypes or OTUs meeting this threshold! Only 24 modules (<10%) differentially covered by body site Compare with 168 modules (>66%) differentially abundant by body site 16 (of 251) modules strongly “core” at 90%+ coverage in 90%+ individuals at 7 body sites 24 modules at 33%+ coverage 71 modules (28%) weakly “core” at 33%+ coverage in 66%+ individuals at 6+ body sites Contrast zero phylotypes or OTUs meeting this threshold! Only 24 modules (<10%) differentially covered by body site Compare with 168 modules (>66%) differentially abundant by body site

15. Proteoglycan degradation by the gut microbiota 15 AA core Glycosaminoglycans (Polysaccharide chains)

16. Proteoglycan degradation: From pathways to enzymes 16 10 -3 10 -8 Enzyme relative abundance Heparan sulfate degradation missing due to the absence of heparanase, a eukaryotic enzyme Other pathways not bottlenecked by individual genes HUMAnN links microbiome-wide pathway reconstructions → site-specific pathways → individual gene families

17. Patterns of variation in human microbiome function by niche 17

18. Patterns of variation in human microbiome function by niche 18 Three main axes of variation Eukaryotic exterior Low-diversity vaginal Gut metabolism Oral vs. tooth hard surface Only broad patterns: every human-associated habitat is functionally distinct!

19. Normal varies a lot at the genus level (16S) 200 subjects Bacteroides Alistipes Faecalibacterium Parabacteroides 343 genera Relative frequency Relative frequency of genera within Stool Dirk Gevers

20. Bacteroides vulgatus Bacteroides sp. Bacteroides uniformis Bacteroides sp. Bacteroides stercoris Bacteroides caccae Relative frequency of Bacteroides species within Stool 123 samples Relative frequency Normal varies a lot at the species level (WGS) Dirk Gevers

21. What’s wrong with this picture? 21 Lactobacillus crispatus MV-1A-US Lactobacillus crispatus JV-V01 Lactobacillus crispatus 125-2-CHN Lactobacillus crispatus 214-1 Lactobacillus crispatus MV-3A-US Lactobacillus crispatus ST1 Lactobacillus gasseri JV-V03 Lactobacillus gasseri 202-4 Lactobacillus gasseri 224-1 Lactobacillus gasseri MV-22 Bifidobacterium breve DSM 20213 Bifidobacterium dentium ATCC 27679 Mycoplasma hominis Clostridiales genomosp BVAB3 str UPII9-5 Clostridiales genomosp BVAB3 UPII9-5 Gardnerella vaginalis AMD Prevotella timonensis CRIS 5C-B1 Megasphaera genomosp type 1 str 28L Porphyromonas uenonis 60-3 Gardnerella vaginalis 409-05 Gardnerella vaginalis 5-1 Atopobium vaginae DSM 15829 Gardnerella vaginalis ATCC 14019 Lactobacillus jensenii 1153 Lactobacillus jensenii 269-3 Lactobacillus jensenii SJ-7A-US Lactobacillus jensenii 208-1 Lactobacillus jensenii JV-V16 Lactobacillus jensenii 27-2-CHN Lactobacillus jensenii 115-3-CHN Lactobacillus iners AB-1 Lactobacillus iners DSM 13335 52 posterior fornix microbiomes → Species and strains matter – but so does your method for identifying them in a community!

22. Core gene families 22 Gene X is a core gene for Clade Y All subclades of Clade Y must have Gene X as core gene (strict definition) Gene X may be a core gene of several (unrelated) clades We have to relax the definition for taking into account: Low-level gene losses Sequencing errors Gene calls errors Gene X A core gene is a gene strongly conserved within a clade

23. Examples of core genes 23

24. Clade-specific marker genes 24 Gene X Gene X is a marker gene (for Clade Y) if X is a core gene for Y and X never appears outside Clade Y

25. Examples of marker genes 25

26. The BactoChip: high-throughput microbial species identification 26 With Olivier Jousson, Annalisa Ballarini

27. BactoChip: detecting single species 27 With Olivier Jousson, Annalisa Ballarini

28. MetaPhlAn: inferring microbial abundances from metagenomic data using marker genes 28 Map metagenomic reads to marker genes to infer microbial abundances –Normalizing for copy number, gene length, etc. Much faster than existing approaches as the marker gene database is ~50 times smaller than the whole microbial sequence DB  Few hours instead of weeks for Illumina samples with 100Gb of sequence data MetaPhlAn: Metagenomic Phylogenetic Analysis http://huttenhower.sph.harvard.edu/metaphlan

29. MetaPhlAn: synthetic validation on log- normal abundances 29 Summary of 8 synthetic communities composed by 2M reads coming from 200 organisms with log-normal distributed abundances concentrations Species-levelClass-level Species levelClass level

30. Matching 16S and more 30

31. The human microbiome at species-level resolution 31

32. Whence enterotypes? 32 Genera Species

33. Microbial community function and structure in the human microbiome: the story so far? Who’s there varies even in health –What they’re doing doesn’t (as much) –Both correlate with niche –By the way: both change during disease and treatment There are patterns in this variation –Function correlates with membership and phenotype –“Pathogenicity” correlates with lower prevalence –Membership means species, strains, or variants –Patterns aren’t always as simple as enterotypes ~1/3 to 2/3 of human metagenome characterized –Job security! 33

34. Ask both what you can do for your microbiome and what your microbiome can do for you

35. Wendy Garrett Michelle Rooks Ramnik Xavier Harry Sokol Thanks! 35 Nicola SegataLevi Waldron Fah Sathira Human Microbiome Project HMP Metabolic Reconstruction Owen White George Weinstock Karen Nelson Joe Petrosino Mihai Pop Pat Schloss Makedonka Mitreva Erica Sodergren Vivien Bonazzi Jane Peterson Lita Proctor Sahar Abubucker Yuzhen Ye Beltran Rodriguez-Mueller Jeremy Zucker Qiandong Zeng Mathangi Thiagarajan Brandi Cantarel Maria Rivera Barbara Methe Bill Klimke Daniel Haft Dirk Gevers Bruce BirrenMark Daly Doyle WardEric Alm Ashlee EarlLisa Cosimi http://huttenhower.sph.harvard.edu Joseph Moon Vagheesh Narasimhan Tim Tickle Xochi Morgan Josh Reyes Jeroen Raes Karoline Faust Jacques Izard Olivier Jousson Annalisa Ballarini

37. Linking function to community composition 37 ← Taxa and correlated metabolic pathways → ← 52 posterior fornix microbiomes → F-type ATPase, THF Sugar transport Phosphate and peptide transport AA and small molecule biosynthesis Embden-Meyerhof glycolysis, phosphotransferases Eukaryotic pathways Plus ubiquitous pathways: transcription, translation, cell wall, portions of central carbon metabolism… Lactobacillus crispatus Lactobacillus jensenii Lactobacillus gasseri Lactobacillus iners Gardnerella/Atopobium Candida/Bifidobacterium

38. Linking communities to host phenotype 38 Normalized relative abundance Vaginal pH (posterior fornix) Body Mass Index Top correlates with BMI in stool Vaginal pH, community metabolism, and community composition represent a strong, direct link between phenotype and function in these data. Vaginal pH (posterior fornix)