Whole Genome based Phylogeny

Transcription

1 Whole Genome based Phylogeny Johanne Ahrenfeldt PhD student DTU Bioinformatics

2 Short about me Johanne Ahrenfeldt PhD student at DTU Bioinformatics Whole Genome based Phylogeny Graduate Engineer in Systems Biology and Bioinformatics from Technical University of Denmark Working in the CGE project since 2012 started as a student helper

3 Overview What is Phylogeny SNP methods CSI Phylogeny Nucleotide Differences NDtree Controlled Evolution study Good advice A cool new thing called Evergreen

4 What is phylogeny? Early phylogeny Classification Based on phenotypes Current phylogeny Based on genotypes DNA mutations as basis for evolution

5 Classification Carl Linnaeus Hierarchical system Kingdom Phylum Class Order Family Genus Species

6 Classification depicted as a tree

7 Classification depicted as a tree Species Genus Family Order Class

8 DNA mutations as basis for evolution

9 What are phylogenetic trees Phylogenetic trees are a visual representation of the genetic relationship between species Think of them as family trees Phylogeny can also be represented by distance matrices

10 What are phylogenetic trees Trees were traditionally made using aligned sequences of single genes or proteins Whole genome data can be used to create trees based on SNP calling K-mer overlap Alignment of genomes

11 What is a SNP A Single Nucleotide Polymorphism (SNP) is a DNA sequence variation occurring commonly* within a population (e.g. 1%) in which a Single Nucleotide A, T, C or G in the genome (or other shared sequence) differs between members of a biological species or paired chromosomes.

12 How does it work Strain A ATTCAGTAGT Strain B ATGCAGTTGA Strain C ATGCAATTGT Strain D ATCCATTAGC

13 Construct distance matrix Strain A ATTCAGTAGT Strain B ATGCAGTTGA Strain C ATGCAATTGT Strain D ATCCATTAGC A B C D AA B C D

14 Make Tree Strain A Strain B Strain C Strain D ATTCAGTAGT ATGCAGTTGA ATGCAATTGT ATCCATTAGC A B C D A B C D A B C 1 D

15 How to read phylogenetic trees 16 The lineages leading to contemporary species have all b T. Ryan Gregory. Understanding Evolu<onary Trees. Evo Edu Outreach (2008) 1: DOI /s x 15

16 How to read phylogenetic trees T. Ryan Gregory. Understanding Evolu<onary Trees. Evo Edu Outreach (2008) 1: DOI /s x 16

17 What is phylogeny used for Classify taxonomy The classic use Outbreak detection Increasing with WGS data

18 What is phylogeny used for Cholera outbreak in Haiti 2010 Listeria outbreak 2014 Whole-genome Sequencing Used to Investigate a Nationwide Outbreak of Listeriosis Caused by Ready-to-eat Delicatessen Meat, Denmark, Kvistholm Jensen et al. Clin Infect Dis. (2016) 63 (1): doi: / cid/ciw192

19 Case story Vibrio Cholerae outbreak in Haiti followed the 2010 earthquake Rumors said that the outbreak may have come from Nepal, travelling along with UN soldiers from Nepal No proof had been given of this until the Hendriksen et al. paper in 2011 Popula<on Gene<cs of Vibrio cholerae from Nepal in 2010: Evidence on the Origin of the Hai<an Outbreak. Hendriksen et al. 23 August 2011 mbio vol. 2 no. 4 e doi: /mBio

20 Case story Data 24 recent V. cholerae strains from Nepal 10 previously sequenced V. cholerae isolates, including 3 from the Haitian outbreak Analysis Antimicrobial susceptibility testing PFGE (pulsed-field gel electrophoresis) to analyze for genetic relatedness Whole genome sequencing, SNP identification and phylogenetic analysis

21 Case story - Results Resistance profile Suscep5ble Decreased suscep5bility Nepalese strains Hendriksen et al Resistant Tetracycline Ciprofloxacin Trimethoprim, Sulfamethoxazole Nalidixic Hai<an outbreak strains Centers for Disease Control and Preven<on, 2010 Tetracycline Ciprofloxacin Trimethoprim, Sulfamethoxazole Nalidixic

22 Case story - Results Pulsed-field gel electrophoresis (PFG)E Nepalese isolates divided in 4 groups Most common Haitian type in same group as four Nepalese strains

23 Case story - Results

24 10 minutes break!

25 snptree First online webserver for constructing phylogenetic trees based on whole genome sequencing snptree- - a web- server to iden<fy and construct SNP trees from whole genome sequence data. Leekitcharoenphon P, Kaas RS, Thomsen MC, Friis C, Rasmussen S, Aarestrup FM. BMC Genomics. 2012;13 Suppl 7:S6.

26 snptree flow

27 CSI Phylogeny SNP identification same as snptree Strict sorting of SNPs Depth Relative depth Distance between SNPs SNP quality Read mapping quality Rolf S. Kaas, Pimlapas Leekitcharoenphon, Frank M. Aarestrup, Ole Lund. Solving the Problem of Comparing Whole Bacterial Genomes across Different Sequencing Plaeorms. PLoS ONE 2014; 9(8): e

28 CSI Phylogeny Requires all SNPs to be significant Z-score higher than 1.96 for all SNPs Z = X Y X+Y X is the number of reads, with the most common nucleotide at that position, and Y the number of reads with any other nucleotide.

29 CSI Phylogeny Output Tree build by FastTree algorithm, in Newick format Branch lengths is substitutions per site at the variable sites Matrix of SNP pair counts in text (.txt) format Diagonal SNP matrix

30 CSI Phylogeny

31 NDtree Nucleotide calling A different approach where the main distinction is not between if a SNP should be called or not, but between whether or not there is solid evidence for the nucleotide at the given position. Real- Time Whole- Genome Sequencing for Rou<ne Typing, Surveillance, and Outbreak Detec<on of Verotoxigenic Escherichia coli. Joensen KG, Scheutz F, Lund O, Hasman H, Kaas RS, Nielsen EM, Aarestrup FM. J Clin Microbiol May;52(5):

32 NDtree Simple mapping approach Cuts all reads into K-mers Maps all K-mers to reference genome Makes an ungapped consensus sequences of equal lengths

33 Mapping K- mers Consensus sequence Reference genome Reference genome Genome 1 Genome 2 Genome 3 Genome 4 Genome 5 Genome 6 33

34 NDtree Nucleotide calling When all reads have been mapped the significance of the base call at each position was evaluated by calculating the number of reads X having the most common nucleotide at that position, and the number of reads Y supporting other nucleotides. A Z-score threshold is calculated Z = X Y X+Y > 1.96 (or 3.29) >90% of reads supporting the same base

35 NDtree Count nucleotide differences Method 1: Each pair of sequences was compared and the number of nucleotide differences in positions called in all sequences was counted. More accurate (Z=1.96 is used as threshold) Method 2: Each pair of sequences was compared and the number of nucleotide differences in positions called in both sequences was counted. More robust (Z=3.29 is used as threshold)

36 Method 1 all called Significant posi<ons in Genome 1 Significant posi<ons in Genome 2 Significant posi<ons in Genome 3 Posi<ons used for phylogeny Method 2 pairwise called Significant posi<ons in Genome 1 Significant posi<ons in Genome 2 Significant posi<ons in Genome 3 Posi<ons used between 1 and 2 Posi<ons used between 1 and 3 Posi<ons used between 2 and 3

37 NDtree Uses two different algorithms to make two different trees UPGMA Neighbor Joining Both algorithms are part of the PHYLIP Neighbor program package and make trees from distance matrices

38 UPGMA vs. Neighbor Joining UPGMA works when samples have been taken the same time Neighbor Joining is better when samples have been taken at different times

39 NDtree Output distance.txt: Distance matrix - tab separated dist.mat: Distance matrix - PHYLIP format tree.nj.newick: Neighbor Joining tree - Newick format Branch lengths is number of Nucleotide Differences tree.upgma.newick: UPGMA tree Newick format Branch lengths is number of Nucleotide Differences

40 Controlled Evolution study Single$ colonies$of$ CSH114$ Choose$colony$ Grow$for$8$h$$ Plate$out$ Grow$for$16$h$ Choose$colonies$ Grow$for$8$h$ Plate$out$ Grow$for$16$h$ Choose$colonies$ Grow$for$8$h$ 128x$ Day$1$ Day$2$ Day$3$ $$$$$ $$$$$$$$$$$$$$Day$8$$ For$each$8$hour$culture$a$sample$was$saved$for$DNA$sequencing$ J. Ahrenfeldt, C. Skaarup, H. Hasman, A. G. Pedersen, F. M. Aarestrup and O. Lund. Bacterial whole genome- based phylogeny: construc<on of a new benchmarking dataset and assessment of some exis<ng methods. BMC Genomics (2017) 18:19

41 Naming the descendants Day$1 Day$2 Day$3 Day$4 Day$5 S1111 S111 S1112 S11 S1121 S112 S1122 S1 S1211 S121 S1212 S12 S1221 S122 S1222 S S2111 S211 S2112 S21 S2121 S212 S2122 S2 S2211 S221 S2212 S22 S2221 S222 S2222

42 Mutations

43 Phylogenetic tree using NDtree (UPGMA) S1 S11 S112 S1121 S1122 S111 S1111 S1112 S1212 S121 S1211 S1221 S12 S1222 S122 S S2 S2211 S221 S2212 S22 S2221 S222 S2111 S211 S2112 S21 S212 S2121 S2122

44 Phylogenetic tree using NDtree (Neighbor Joining) S S11 S111 S1111 S1112 S112 S1121 S1122 S1212 S121 S1211 S1222 S1221 S122 S12 S1 S2111 S211 S2112 S212 S2121 S2122 S21 S2211 S2212 S221 S2221 S222 S22 S2

45 UPGMA vs. Neighbor Joining UPGMA works when samples have been taken the same time Neighbor Joining is better when samples have been taken at different times

46 CSI Phylogeny Default settings 1_1_1_1 1_1_1_2 1_1_1 1_1_2_1 1_1_2_2 1_1_2 1_1 1_2_1 1_2_2_1 1_2_2_2 1_2_2 1_2 1 c_1_1_1_2_1 e_1_1_2_1_1 f_1_1_2_1_2 f_2_1_1 g_1_1_2_2_1 h_1_1_2_2_2 m_1_2_2_1_1 n_1_2_2_1_2 o_1_2_2_2_1 d_1_1_1_2_2 b_1_1_1_1_2 1_2_1_1 i_1_2_1_1_1 1_2_1_2 k_1_2_1_2_1 l_1_2_1_2_2 a_1_1_1_1_1 j_1_2_1_1_2 b_1_2_2 0.2

47 CSI Phylogeny Pruning disabled S S1 S11 S112 S2 S121 S1211 S1212 S12 S122 S1221 S1222 S1112 S1111 S111 S1121 S1122 S21 S2111 S211 S2112 S2122 S2121 S212 S221 S2211 S2212 S2221 S222 S22 0.2

48 So What should I use when? CSI Phylogeny Has very good statistics and a good graphical overview. Advantageous to use when you expect the differences between the isolates to be larger than 5-10 mutations. Is faster NDtree Is able to find very small differences. Does not take recombination into consideration. Works best on raw reads. If given assembled genomes, it simulates reads.

49 Choosing a reference genome For comparison of very closely related isolates, a better level of detail is given by using a closely related reference genome.

50 What defines an outbreak We can t tell for certain It depends on the species But a rule of thump is: Within 10 SNPs it is definitely an outbreak Within 30 SNPs it might be an outbreak Above 60 SNPs it is most likely not an outbreak

51 A.C. Schürch et al. / Clinical Microbiology and Infection xxx (2018) 1e5 3 Table 1 Examples of relatedness criteria for wg/cgmlst and SNP typing schemes of representative clinically relevant bacteria Organism Relatedness threshold a References wg/cgmlst (allele) SNPs Acinetobacter baumannii 8 3 [25,26] Brucella spp. Epidemiologic validation in progress b Campylobacter coli, C. jejuni [27,28] Cronobacter spp. Epidemiologic validation in progress b Clostridium difficile Epidemiologic validation in progress b 4 [29], Enterococcus faecium [30] Enterococcus raffinosus Epidemiologic validation in progress b Escherichia coli [31,32], Francisella tularensis 1 2 [33,34] Klebsiella oxytoca Epidemiologic validation in progress b Klebsiella pneumonia [35,36] Legionella pneumophila 4 15 [37] Listeria monocytogenes 10 3 [38,39] Mycobacterium abscessus 30 [40] Mycobacterium tuberculosis [41] Neisseria gonorrhoeae Epidemiologic validation in progress b 14 [42], Neisseria meningitidis Epidemiologic validation in progress b Pseudomonas aeruginosa [31,43] Salmonella dublin Epidemiologic validation in progress b 13 [44], Salmonella enterica Epidemiologic validation in progress b 4 [45], Salmonella typhimurium Epidemiologic validation in progress b 2 [46], Staphylococcus aureus [47,48] Streptococcus suis 21 [49] Vibrio parahaemolyticus 10 [50] Yersinia spp. 0 [51] cg, core genome; MLST, multilocus sequence typing; SNP, single nucleotide polymorphism; wg, whole genome. a Data often represent single studies that can be used to begin formulation of species-specific interpretation criteria. Thus, these data should be coupled with newly published similar studies to ensure that resulting values are not atypical and can be generally applied. b Proposed wg/cgmlst schemes are available online ( but as yet have not been epidemiologically validated.

52 A cool new thing called Evergreen Every 24 hours For every sample Databases Download all new data from SRA/ENA Typing Find closest match in homology reduced database NCBI Infer/recalculate phylogenetic tree For each reference genome with new mapped samples. Infer tree from cluster representatives and add the redundant isolates to their respective nodes based on the clustering Map to closest match Find the phylogenetic distance to consensus sequences for the given reference. Cluster isolates with < 10 SNPs distance Hobohm 1 homology reduction on new, non-clustered isolates Homology reduced database Hobohm 1 clustering to 99% homology on all bacterial reference genomes from NCBI Database of previous homology reduced isolates (representative consensus sequence) and redundant isolates

53 Downloaded and identified by March 9 th 2018 Species Total Included in a tree Salmonella enterica Escherichia coli Listeria monocytogenes Campylobacter jejuni Shigella sonnei Shigella 8lexneri Shigella boydii Campylobacter coli Salmonella bongori 9 9 Pseudomonas aeruginosa 7 3 Escherichia albertii 6 5 Hafnia alvei 4 3 Klebsiella pneumoniae 4 3 Bacillus subtilis 3 3 Bacillus thuringiensis 3 0 Providencia stuartii 2 2 Enterobacter cloacae 2 1 Proteus mirabilis 2 2 Streptococcus agalactiae 1 1 Peptoclostridium dif=icile 1 1 Staphylococcus epidermidis 1 1 Leuconostoc citreum 1 0 Streptococcus pasteurianus 1 0 Shigella dysenteriae 1 1 Raoultella ornithinolytica 1 1 In total

54

55 Thank you for listening Questions?