- PDF Free Download

Size: px

Start display at page:

Download ""

Scott Gaines
5 years ago
Views:

1 choose MBL-REGISTER user: dm00834 password: dm

2 stamps.mbl.edu this uses the username and password on your STAMPS name badge

6 Strategies for Analysis of Microbial Population Structures Tag/Marker Gene Metagenomics (sequencing amplified markers, amplicons e.g. SSU rrna) Who s in the community? Shotgun Metagenomics (sequencing total community DNA) What is the functional potential of the community? (Partial) Assembly of community genomes Metatranscriptomics (sequencing tag or shotgun cdna) What fraction of the community is active? What is the community doing? Metaproteomics (sequencing community peptides) Meta-metabolomics (detecting and quantifying metabolites used by the community) Total Community Sampling vs Targeted or Enriched Sampling

7 Tag/Marker Gene Sequencing Metagenomics 0. Design your experiment to meet your goals! - Sampling depth (more reads/sample) vs. sampling breath (more samples) - Targeted approach vs trying to sample everything - Importance of rare taxa? - Importance of biological replicates - Collect appropriate contextual metadata - Standardize experimental approach before you begin

8 Tag/Marker Gene Sequencing Metagenomics 1. Design primers to a region of interest - Usually SSU rrna ( 16S ) - conserved primers flanking variable regions(s) - length dependent on sequencing platform 2. Extract DNA (RNA) (2a Create cdna from RNA) 3. Amplify DNA All of these steps introduce biases! 4. Sequence Also biased, but probably less so than the earlier steps!

9 Tag/Marker Gene Sequencing Metagenomics 5. Clean up the data - Remove low quality reads - Remove other error-prone reads - Remove chimeric reads - Remove non-target reads 6. Organize the reads into units Robert Edgar, Today Meren, Wednesday Susan Holmes, Wed Rob Knight, Thursday Tracy Teal, Friday - Assign reads to a taxonomic identifier - Assign reads to a phylogenetic clade - Group reads into (or assign reads to) Operational Taxonomic Units (OTUs) By sequence similarity By sequence information

10 Tag/Marker Gene Sequencing Metagenomics 7. Analyze your groups of sequences - Assumption: each read represents an individual - Assumption: steps 1-6 are effectively unbiased, or equally biased across samples - Quandary: what to do about low abundant groups (singletons)? - Quandary: how to compare samples of unequal size? Within a sample or combined samples - Richness, number of observed and estimated groups - Evenness, relative abundance of the groups Between samples or groups of samples - Beta diversity - Relating diversity to metadata Amy Willis, Tues Robert Edgar, Tues Meren, Wednesday Susan Holmes, Wed Rob Knight, Thurs Tracy Teal, Friday

11 An Abridged History of DNA Sequencing 1971 Wu & Taylor sequence overhang of phage λ (12 bp) 1974 Sogin, Woese and Pace sequence a 5S rrna (116 nt) 1977 Sanger et al. sequence phix174 (5,375bp) 1982 Sanger et al. sequence phage λ (48,501bp) 1985 Pace, Olsen, Sogin, and others sequence 16S rrna genes 1996 ABI Prism 310 Genetic Analyzer 1998 Phred automated base calling becomes standard 1998 ABI 3700; high-throughput capillary sequencing becomes widely available GS Solexa Genome Analyzer 2011 Illumina HiSeq + PacBio, Nanopore, 10X, etc...

12 Cost of DNA Sequencing $10, $1, : $10,000/bp (first sequencing) 1977: $10/bp (first Sanger sequencing) 1996: $1/bp (ABI 310) 1998: $0.1/bp (ABI 3700) 2001: $0.001/bp (ABI3730) Cost per Mb $ $10.00 $1.00 $0.10 $0.01 Sep-01 Mar-02 Sep-02 Mar-03 Sep-03 Mar-04 Sep-04 Mar-05 Sep-05 Mar-06 Sep-06 Mar-07 Sep-07 Wetterstrand KA. DNA Sequencing Costs: Data from the NHGRI Genome Sequencing Program (GSP) Available at: Mar-08 Sep-08 Mar-09 Sep-09 Mar-10 Sep-10 Mar-11 Sep-11 Mar-12 Sep-12 Mar-13 Sep-13 Mar-14 Sep-14 Mar-15

13 Q: To what degree does your research drive nextgen sequencing technology?

14 The Pattern of Perception true acceptance performance The Collins Curve disillusionment initial idea technology 1 technology 2 technology 3

15 Sanger Capillary ABI Sequencing 96 reactions 48 clones 30.5 MB 317K M10MaGB08.b.ab1 317K M10MaGB08.g.ab1 ABI 3700 chromatogram

16 Phred Quality Scores Q = -10*log(P) Q P in in in 1, in 10,000 In the era of high-throughput capillary sequencing, the gold standard to which next-gen methods are measured, the goal was to produce reads with an average phred score of 20.

Illumina HiSeq/MiSeq/NextSeq 275M KEG1_ACTTGA_L004_R1_001.fastq.gz 288M KEG1_ACTTGA_L004_R1_002.fastq.gz 277M KEG1_ACTTGA_L004_R1_003.fastq.gz 286M KEG1_ACTTGA_L004_R1_004.fastq.gz 279M KEG1_ACTTGA_L004_R1_005.

17 Illumina HiSeq/MiSeq/NextSeq 275M KEG1_ACTTGA_L004_R1_001.fastq.gz 288M KEG1_ACTTGA_L004_R1_002.fastq.gz 277M KEG1_ACTTGA_L004_R1_003.fastq.gz 286M KEG1_ACTTGA_L004_R1_004.fastq.gz 279M KEG1_ACTTGA_L004_R1_005.fastq.gz 286M KEG1_ACTTGA_L004_R1_006.fastq.gz 279M KEG1_ACTTGA_L004_R1_007.fastq.gz 283M KEG1_ACTTGA_L004_R1_008.fastq.gz 279M KEG1_ACTTGA_L004_R1_009.fastq.gz 281M KEG1_ACTTGA_L004_R2_001.fastq.gz 293M KEG1_ACTTGA_L004_R2_002.fastq.gz 283M KEG1_ACTTGA_L004_R2_003.fastq.gz 292M KEG1_ACTTGA_L004_R2_004.fastq.gz 285M KEG1_ACTTGA_L004_R2_005.fastq.gz 291M KEG1_ACTTGA_L004_R2_006.fastq.gz 285M KEG1_ACTTGA_L004_R2_007.fastq.gz 287M KEG1_ACTTGA_L004_R2_008.fastq.gz 286M KEG1_ACTTGA_L004_R2_009.fastq.gz 75,000,000 reads 375,000,000 reads per lane 5.1GB of data... from one lane

18 fastq 2:N:0:ACTTGA 2:N:0:ACTTGA ATTCATGTCGCTGATGAAAGGCGCTGGAGCCAACATTCTCAGAGGCATTGCTGGCGCTGGTGTCCTATCAGGCTTCGACAAGCT + 2:N:0:ACTTGA TTGTAAACTTTGAAATTAAAACTCAAAATGGGTAGCCTTTATCAGAGTTCTATTCTCAATTCTATTTTAAAATGGACCAAGCTC + 2:N:0:ACTTGA GAGCCCAATTCCGCAGAAACTTGCCATCAAAGGCCATCGGAAGAAGAGTGCGAGTTATGTTGTATCCTTCAAGTATTTAAAAAG + 2:N:0:ACTTGA GTTTGATTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTAAAATTCCAAATTTACCG + 2:N:0:ACTTGA ACTAATAGAATACTTTAACTGCCTCTACGTGAGAGTGGAAAGCAATAGGAAGGCGGGGTGATCACTCTAAATGTACAGAATTAT + 2:N:0:ACTTGA

19 fastq format machine run flowcell lane tile pair (1, 2, 0) x:y coordinates failed filter? multiplex 2:N:0:ACTTGA CAACTGCAGACGTTTGGTCCCAAAAGATAAAGCGATCAAAAAATTCATTATCAGAAATATTGTCGAGGCTGCAGCTGTCAGAGA Phred quality score from 0 to 93 using ASCII 33 to ! # $ A B C D ~ }

20 Tag Sequencing Metagenomics 7. Analyze your groups of sequences - Assumption: each read represents an individual - Assumption: steps 1-6 are effectively unbiased, or equally biased across samples - Quandary: what to do about low abundant groups (singletons)? - Quandary: how to compare samples of unequal size? Within a sample or combined samples - Richness, number of observed and estimated groups - Evenness, relative abundance of the groups Between samples or groups of samples - Beta diversity - Relating diversity to metadata

21 Operational Taxonomic Units (OTUs) one read one sequence five reads one sequence 26 reads 14 sequences 5 OTUs ten reads one sequence Singleton OTU

22 Rank Abundance Curve 100,000 one OTU has 647,903 tags tags / OTU 10,000 1, observed evenness observed richness ,000 20,000 10,000 40,000 50,000 60,000 80,664 OTUs have one tag each 70,000 80,000 90, , , , ,000 OTUs Ranked by Abundance

23 tags/otu Number of OTUs Species Abundance Curve

24 Species Abundance Curve How many OTUs have zero tags each (missing)? 80,664 OTUs have one tag each (singletons) 18,817 OTUs have two tags each (doubletons) Number of OTUs Total obs OTUs + missing OTUs = Total alpha diversity 160, ,609 = 637,763 1 OTU has 1,993 tags one OTU has 647,903 tags tags/otu

25 Rarefaction Curve: What is it? OTUs Observed Tags Sampled

26 Rarefaction Curve: what is it? Species Abundance Curve original sample has 78,650 tags in 3,320 OTUs Number of OTUs tags/otu

27 Rarefaction Curve: what is it? Species Abundance Curve original sample has 78,650 tags in 3,320 OTUs create pseudo-replicate sample with n tags by removing 78,650 n tags from the SA curve count how many OTUs have at least one tag remove 60,650 tags 18,000 tags remain 1,604 OTUs still represented repeat for n = 100, 200, or other interval

28 Rarefaction Curve OTUs Observed Many low-abundance OTUs always missing in pseudo replicates High abundance OTUs always present in pseudo replicates Tags Sampled

29 Rarefaction Curve OTUs Observed Same tag data, but: sub-samples of tags clustered into OTUs rarefaction run on these new OTUs Tags Sampled

30 OTUs Observed 2500 Rarefaction Curve Same tag data, but: sub-samples of tags clustered into OTUs rarefaction run on these new OTUs This should not happen! The solution is not to sub-sample to some common tag count to avoid the problem! Tags Sampled

31 Rarefaction Curve OTUs Observed This sub-sampling phenomenon indicated something was not happening the way it was supposed to happen. There was a problem with the procedure. An improved clustering method generates OTUs that do not display the phenomenon. Understand what the methods do so that you can recognize when there is a problem! Tags Sampled

32 Tag Sequencing Metagenomics 7. Analyze your groups of sequences - Assumption: each read represents an individual - Assumption: steps 1-6 are effectively unbiased, or equally biased across samples - Quandary: what to do about low abundant groups (singletons)? - Quandary: how to compare samples of unequal size? Within a sample or combined samples - Richness, number of observed and estimated groups - Evenness, relative abundance of the groups Between samples or groups of samples - Beta diversity - Relating diversity to metadata

33 International Census of Marine Life million bacterial sequence tags from 549 surveys

34 Species Abundance of 3% OTUs from the ICoMM Survey Number of OTUS % singletons! Global survey of crustaceans across 93 coral reef sites... 38% singletons! tags/otu Bacteria otus.icomm.sabund

total dataset 1 dataset 2 dataset 3 dataset 4 dataset 5 OTU-1 49 12 14 10 11 2 OTU-3 26 10 1 0 7 8 OTU-4 12 1 3 3 4 1 OTU-2 10 5 3 1

35 total dataset 1 dataset 2 dataset 3 dataset 4 dataset 5 OTU OTU OTU OTU OTU Singleton in dataset 1 but not in total Singleton in dataset 3 not present in any other dataset unique singleton

36 Richness Distributions of 100 Most Abundant OTUs OTUs Bacteria otusizes.icomm

37 The most abundant OTUs are rare somewhere If an OTU appears with abundance greater than 100 in one dataset, there is a ~90% chance it appears as a singleton in another Most singletons in a dataset are found elsewhere 79% of the singleton OTUs in a dataset are not unique

38 What is a singleton? ~1 x 10 9 bacterial cells/liter epipelagic seawater Based on average ICoMM dataset abundances: 1-3 OTUs represent >1% each, or more than 100,000,000 cells/l ~10 OTUs represent 0.1% 1%, or 10, ,000,000 cells/l ~100 OTUs represent %, or 1,000 10,000,000 cells/l ~1,000 OTUs represent less than 0.001%, that s still ~ 100,000 cells/l

Cells/Liter 10 8 10 7 10 6 10 5 Our Sampling is Very Incomplete, even

39 Cells/Liter Our Sampling is Very Incomplete, even with 20,000 Tags / Liter 10 4 Tags/Dataset Rank Abundance

40 P p (n N) = p = cells/ml n = 1 N = 20,000 What is a Singleton? If sampling 20,000 tags from 1 L at 10 9 cells/l, a singleton likely represents an OTU with an abundance of 50, ,000 cells/l P p (n N) = N! n!(n-n)! pn (1-p) N-n Cells/OTU/Liter

41 Is there a real difference in abundance between an OTU observed as singleton and as a doubleton? Singleton Doubleton Cells/OTU/Liter

42 Is there a real difference in abundance between an OTU observed as singleton and as a doubleton? Cells/OTU/Liter

43 What s Represented in 20,000 tags? (assuming sampling 20,000 tags from 1 liter at 10 9 cells/liter) singletons doubletons Missing! Cells/OTU/ml

44 Understand the methods you are using; recognize when the answers don t make sense Appreciate the biology of your system; recognize when your results are in conflict and be suspicious of your results when this happens Document everything you do

Sequencing Errors, Diversity Estimates, and the Rare Biosphere

Sequencing Errors, Diversity Estimates, and the Rare Biosphere or Living in the shadow of Errares Susan Huse Marine Biological Laboratory June 13, 2012 Consistent Community Profile across samples and environments