toward understanding our microbial world. J. Gregory Caporaso Department of Biological Sciences, Northern Arizona University TwiLer/GitHub: @gregcaporaso www.caporasolab.us Photo by John Spear
Back of the envelope calculaQon: less than 13% of bacterial species* have a representaQve that has been grown in culture. Many recent advances are based on culture-‐independent approaches for studying microbial communiQes. * Defined as 97% OTUs in the Greengenes 13_5 reference database.
life has a shared evoluQonary history, and some genes are shared by all organisms. The sequence of those genes can be used as a gene3c fingerprint for different organisms.
ACTAGCAT ACTAGTAT TCAATGTT TCCATGTT ACCAGGTT Time The random accumulaQon of muta3ons (changes to gene sequences over evoluQonary Qme) gives us informaQon for idenQfying and comparing organisms. Escherichia Desulfovibrio Thermus Thermoplasma Haloferax
to “fingerprint” different microbial organisms. Why this gene? • It’s ubiquitous. • Contains regions that idenQcal across organisms, and regions that are variable across organisms.
samples using barcoded PCR, and sequence. Pool samples and sequence Micah Hamady, et al., Nature Methods, 2008. Error-‐correcQng barcodes for pyrosequencing hundreds of samples in mulQplex. Per-‐sample rRNA >GCACCTGAGGACAGGCATGAGGAA… >GCACCTGAGGACAGGGGAGGAGGA… >TCACATGAACCTAGGCAGGACGAA… >CTACCGGAGGACAGGCATGAGGAT… >TCACATGAACCTAGGCAGGAGGAA… >GCACCTGAGGACACGCAGGACGAC… >CTACCGGAGGACAGGCAGGAGGAA… >CTACCGGAGGACACACAGGAGGAA… >GAACCTTCACATAGGCAGGAGGAT… >TCACATGAACCTAGGGGCAAGGAA… >GCACCTGAGGACAGGCAGGAGGAA…
taxons (or OTUs) on a per-‐sample basis Also will generally have taxonomy classificaQon for each taxon, and a phylogeneQc tree providing esQmates of the evoluQonary distance between all taxa.
(discrete data) or area plots (conQnuous data) staQsQcally – currently an acQve area of research Kruskal-‐Wallace, ANOVA are defaults in QIIME, though recent evidence suggests other approaches are beLer: Waste not, want not: why rarefying microbiome data is inadmissible. McMurdie and Holmes, 2014. hLp://www.ncbi.nlm.nih.gov/pubmed/24699258 (Available in QIIME: differential_abundance.py) Analysis of composiQon of microbiomes… Mandal et al., 2015. hLp://www.ncbi.nlm.nih.gov/pubmed/26028277 (Not yet available in QIIME, but coming soon.) Plant Human Dog Turtle Time (not relevant for this example) è (Not real data!)
visually with ordinaQon plots (e.g., PCoA, NMDS) staQsQcally with ANOSIM* * In QIIME 1.9.1, use compare_categories.py Unweighted UniFrac distance matrix: Sample ID Sample Type A Plant (yellow) B Plant (yellow) C Turtle (red) D Turtle (red) E Human (green) F Dog (blue)
(yellow) B Plant (yellow) C Turtle (red) D Turtle (red) E Human (green) F Dog (blue) (Not real data!) Beta diversity comparison: visually with ordinaQon plots (e.g., PCoA, NMDS) staQsQcally with ANOSIM* * In QIIME 1.9.1, use compare_categories.py Unweighted UniFrac distance matrix: OrdinaQon plot:
few of the >3200 arQcles ciQng QIIME (Google Scholar, 30 Sept 2015) Cell (2012) Applied and Environmental Microbiology (2011) Current InfecQous Disease Reports (2011) PLoS One (2011) Microbiology and Molecular Biology Reviews (2013)
153,408-‐core compute cluster*) Virtual machines VirtualBox (local) or Cloud environments (e.g., Amazon Web Services, iPlant) NEW: Now available as a BaseSpace app! *Hopper (hLp://i.top500.org/system/176952) www.qiime.org
• Call any script with –h to get help or see the script usage pages – hLp://qiime.org/documentaQon/script_index.html • Ask quesQons on the QIIME Forum – hLp://forum.qiime.org • Report bugs on the issue tracker – hLp://github.com/qiime/qiime/issues
user-‐defined (tab-‐separated text; we recommend using Google Docs) Sequence files (in fasta, fastq, or sff format) OTU table: sample x OTU matrix, central to downstream analyses (in BIOM format) PhylogeneQc tree (if applicable; in newick format)
(see hLp://bit.ly/mphm-‐qiime ) Required fields for demulQplexing steps: SampleID, BarcodeSequence, LinkerPrimerSequence, DescripQon Provide all of the informaQon you have about your samples – metadata is the key to interpretaQon.
(see hLp://bit.ly/mphm-‐qiime ) More informaQon on metadata: MIMARKS/MIxS standards: hLp://www.nature.com/nbt/journal/v29/n5/full/nbt.1823.html Formawng QIIME mapping files: hLp://qiime.org/documentaQon/file_formats.html Working with Google Docs from QIIME (experimental, and will become more integrated in the future): hLp://qiime.org/tutorials/remote_mapping_files.html For example, to download this mapping file you can run the command: load_remote_mapping_file.py -‐k 0AvglGXLayhG7dGNuQWJKMlNWVFdVXzNlRXYybjFJV2c –o tutorial-‐map.txt
These should be obtained as sQll-‐mulQplexed data from your sequencing center (i.e., not yet mapped from barcode to sample). This allows for use of QIIME’s quality filtering. For more informaQon: Quality filtering Illumina data hLp://www.nature.com/nmeth/journal/v10/n1/full/nmeth.2276.html DemulQplexing 454 data with QIIME hLp://qiime.org/tutorials/tutorial.html Denoising 454 data with QIIME hLp://qiime.org/tutorials/denoising_454_data.html DemulQplexing Illumina data with QIIME hLp://qiime.org/tutorials/processing_illumina_data.html
(.biom) format Call: biom convert -h for converQng between classic and BIOM OTU tables. Detailed usage examples available here: hLp://biom-‐format.org/documentaQon/biom_conversion.html
How I Learned To Stop Worrying and Love the Ome-‐ome JSON-‐based format for represenQng arbitrary sample x feature count tables with opQonal metadata McDonald et al., GigaScience (2012).
Human skin, colored by individual, at 500 sequence/sample Image/analysis credit: JusQn Kuczynski Data reference: Forensic idenQficaQon using skin bacterial communiQes. Fierer N, Lauber CL, Zhou N, McDonald D, Costello EK, Knight R. Proc Natl Acad Sci U S A. 2010 Apr 6;107(14):6477-‐81.
idenQficaQon using skin bacterial communiQes. Fierer N, Lauber CL, Zhou N, McDonald D, Costello EK, Knight R. Proc Natl Acad Sci U S A. 2010 Apr 6;107(14):6477-‐81. VariaQon in sampling depth is an important consideraQon Human skin, colored by sampling depth, at either 50 or 500 sequences/sample
50 (blue) or 500 (red) sequences/sample Image/analysis credit: JusQn Kuczynski Data reference: Forensic idenQficaQon using skin bacterial communiQes. Fierer N, Lauber CL, Zhou N, McDonald D, Costello EK, Knight R. Proc Natl Acad Sci U S A. 2010 Apr 6;107(14):6477-‐81. VariaQon in sampling depth is an important consideraQon
human subjects provided daily microbiome samples, one for 6 months and the other for 18 months. • Sampled tongue, lez and right palms, and gut (via feces). • Tutorial data is a subset of this full data set. • Illumina HiSeq 2000 16S sequencing with Earth Microbiome Project protocols (earthmicrobiome.org)
United States License. To view a copy of this license, visit hLp://creaQvecommons.org/licenses/by/3.0/us/ or send a leLer to CreaQve Commons, 171 Second Street, Suite 300, San Francisco, California, 94105, USA. Feel free to use or modify these slides, but please credit us by placing the following aLribuQon informaQon where you feel that it makes sense: Slides derived from QIIME educa3onal materials www.qiime.org. www.qiime.org Slides compiled by: Greg Caporaso John Chase Jose Clemente Antonio Gonzalez Peña Rob Knight Cathy Lozupone Daniel McDonald Jai Rideout Yoshiki Vázquez Baeza
Jose Clemente John Chase Kevin Cohen Antonio Gonzalez Crystal Hepp Rob Knight Bruce Hungate Larry Hunter Paul Keim ScoL Kelley JusQn Kuczynski Cathy Lozupone Daniel McDonald David Mills Norm Pace Fernando Perez Jai Rideout Egbert Schwartz Karen Schwarzberg Jeffrey Siegel Jesse Stombaugh Yoshiki Vazquez Tony Walters gregcaporaso@gmail.com hLp://caporasolab.us TwiLer/GitHub: @gregcaporaso
PC3 (8.1%) PC1 (24%) PC2 (17%) PC3 (9.7%) 100 sequences/sample 10 sequences/sample 1 sequence/sample Direct sequencing of the human microbiome readily reveals community differences. J Kuczynski et al. Genome Biology (2011). How deep is deep enough?
level using the RDP Classifier (80% confidence) Acknowledgement: Tony Walters Accuracy of classified reads 90 95 100 Full Amplicon 515F/806R (150 base read) 515F/806R (100 base read) 96% accurate! Phylum Class Order Family Genus Species 0 20 40 60 80 100 Full Amplicon 515F/806R (150 base read) 515F/806R (100 base read) Phylum Class Order Family Genus Species
Pseudomonas argenQnensis Pseudomonas flavescens Sample B Pseudomonas aeruginosa Pseudomonas argenQnensis Escherichia coli Alpha diversity Conclusion A = B = C Observed species Sample A 3 Sample B 3 Sample C 3 Sample C Pseudomonas aeruginosa Giardia lamblia Methanobrevibacter smithii
Pseudomonas argenQnensis Pseudomonas flavescens Sample B Pseudomonas aeruginosa Pseudomonas argenQnensis Escherichia coli Sample C Pseudomonas aeruginosa Giardia lamblia Methanobrevibacter smithii PD = 0.35 PD = 0.40 PD = 0.95 < < Sample C is more diverse than sample B, which is more diverse than sample A Alpha diversity
clustered based on similarity to one another. • Reference-‐based – Closed reference: any reads which don’t hit a reference sequence are discarded – Open reference: any reads which don’t hit a reference sequence are clustered de novo hLp://qiime.org/tutorials/otu_picking.html
– You do not have a reference sequence collecQon to cluster against, for example because you're working with an infrequently used marker gene. • You cannot use if: – You are comparing non-‐overlapping amplicons, such as the V2 and the V4 regions of the 16S rRNA. – You working with very large data sets, like a full HiSeq 2000 run. (Technically you can, but it will be really slow.) pick_de_novo_otus.py hLp://qiime.org/tutorials/tutorial.html
– Easily parallelizable – OTUs are defined by high-‐quality, trusted sequences • Cons – Reads that don’t hit reference dataset are excluded, so you can never observe new OTUs pick_closed_reference_otus.py
– You are comparing non-‐overlapping amplicons, such as the V2 and the V4 regions of the 16S rRNA. Your reference sequences must span both of the regions being sequenced. • You cannot use if: – You do not have a reference sequence collecQon to cluster against, for example because you're working with an infrequently used marker gene. pick_closed_reference_otus.py
– You are comparing non-‐overlapping amplicons, such as the V2 and the V4 regions of the 16S rRNA. – You do not have a reference sequence collecQon to cluster against, for example because you're working with an infrequently used marker gene. pick_open_reference_otus.py hLp://qiime.org/tutorials/illumina_overview_tutorial.html hLp://qiime.org/tutorials/open_reference_illumina_processing.html hLp://qiime.org/tutorials/fungal_its_analysis.html
gregcaporaso
dpc
matleenalaakso
kakukoki
kikagaku_0117
tibbelit
loochsorg
birdhiro
badarpacmann
event2020
dscs
signer
pauljervisheath
colly
dougneiner
swwweet
keathley
maltzj
jacobian
trishagee
cassininazir
keavy
addyosmani
kneath
