1. Taxonomic identification and phylogenetic profiling
Nam-phuong Nguyen
Carl R. Woese Institute for Genomic Biology
University of Illinois at Urbana-Champaign
Joint work with Siavash Mirarab, Mihai Pop, and Tandy Warnow

2. Metagenomics Culture-independent method for studying a microbiome
Extract genetic material directly from the environment
Applications to biofuel production, agriculture, human health
Sequencing technology produces millions of short reads from unknown species
Fundamental steps in analysis is identifying taxa of read and estimating a population profile of a sample
Metagenomics is the study of sampling genetic material directly from an environmental sample
Courtesy of Human Microbiome Project

3. Taxonomic Identification and Profiling
Objective: Given a query sequence, identify the taxon (species, genus, family, etc...) of the sequence
Classification problem
Taxonomic profiling
Objective: Given a set of query sequences collected from a sample, estimate the population profile of the sample
Estimation problem
Can be solved via taxonomic identification
The two basic steps in phylogenetic placement is similar to the two phase methods we typically use, align and then place the fragment.
The differences are we align each fragment independently into the backbone tree. We then take each extended alignment and place each fragment independently into the backbone tree.
Two methods to align the query sequences are HMMALIGN, which estimates a hidden markov model on backbone alignment, then aligns query sequence. PaPaRa, which estimates ancestral parsimony state vectors for each in the reference tree and aligns the query sequence against each state vector. It selects the best scoring alignment.
Two methods that place the query sequence try to optimize the same criterion: find the ML placement given the extended alignment

4. Taxonomic Identification Methods
Sequence similarity search
Classifies by finding most similar sequence
Classifies fragments from any region of genome
BLAST
Composition-based methods
Typically uses k-mers
PhymmBL, NBC
Phylogeny-based methods
Classifies fragments by using a phylogeny
The human body contains 10 times more bacterial cells than human cells

5. Phylogeny-based taxonomic identification5
Phylogeny-based taxonomic identification
Known full-length gene sequences,
and an alignment and a tree
Fragmentary Reads:
( bp long)
(500-10,000 bp long)
ACCG
CGAG
CGG
GGCT
TAGA
GGGGG
TCGAG
GGCG
GGG .
ACCT
Remove internal nodes
TAGA...CTT
(species3)
AGC...ACA
(species4)
ACT..TAGA..A
(species5)
AGG...GCAT
(species1)
TAGC...CCA
(species2)

6. Phylogeny-based taxonomic identification6
Phylogeny-based taxonomic identification
Known full-length gene sequences,
and an alignment and a tree
Fragmentary Reads:
( bp long)
(500-10,000 bp long)
ACCG
CGAG
CGG
GGCT
TAGA
GGGGG
TCGAG
GGCG
GGG .
ACCT
Remove internal nodes
TAGA...CTT
(species3)
AGC...ACA
(species4)
ACT..TAGA..A
(species5)
AGG...GCAT
(species1)
TAGC...CCA
(species2)

7. Phylogeny-based taxonomic identification7
Phylogeny-based taxonomic identification
Known full-length gene sequences,
and an alignment and a tree
Fragmentary Reads:
( bp long)
(500-10,000 bp long)
ACCG
CGAG
CGG
GGCT
TAGA
GGGGG
TCGAG
GGCG
GGG .
ACCT
Remove internal nodes
TAGA...CTT
(species3)
AGC...ACA
(species4)
ACT..TAGA..A
(species5)
AGG...GCAT
(species1)
TAGC...CCA
(species2)

8. Phylogeny-based taxonomic identification8
Phylogeny-based taxonomic identification
Known full-length gene sequences,
and an alignment and a tree
Fragmentary Reads:
( bp long)
(500-10,000 bp long)
ACCG
CGAG
CGG
GGCT
TAGA
GGGGG
TCGAG
GGCG
GGG .
ACCT
Remove internal nodes
TAGA...CTT
(species3)
AGC...ACA
(species4)
ACT..TAGA..A
(species5)
AGG...GCAT
(species1)
TAGC...CCA
(species2)

9. Phylogeny-based taxonomic identification9
Phylogeny-based taxonomic identification
Known full-length gene sequences,
and an alignment and a tree
Fragmentary Reads:
( bp long)
(500-10,000 bp long)
ACCG
CGAG
CGG
GGCT
TAGA
GGGGG
TCGAG
GGCG
GGG .
ACCT
Remove internal nodes
TAGA...CTT
(species3)
AGC...ACA
(species4)
ACT..TAGA..A
(species5)
AGG...GCAT
(species1)
TAGC...CCA
(species2)

10. Phylogeny-based taxonomic identification10
Phylogeny-based taxonomic identification
Known full-length gene sequences,
and an alignment and a tree
Fragmentary Reads:
( bp long)
(500-10,000 bp long)
ACCG
CGAG
CGG
GGCT
TAGA
GGGGG
TCGAG
GGCG
GGG .
ACCT
Remove internal nodes
TAGA...CTT
(species3)
AGC...ACA
(species4)
ACT..TAGA..A
(species5)
AGG...GCAT
(species1)
TAGC...CCA
(species2)

11. Phylogenetic Placement
Input: (Backbone) Alignment and tree on full-length sequences and a query sequence (short read)
Output: Placement of the query sequence on the backbone tree
Use placement to infer relationship between query sequence and full-length sequences in backbone tree
Applications in metagenomic analysis
Millions of reads
Reads from different genomes mixed together
Use placement to identify read
Metagenomic analysis, include a slide

12. Align Sequence S1 S2 S4 S3 S1 = -AGGCTATCACCTGACCTCCA-AA
S2 = TAG-CTATCAC--GACCGC--GCA
S3 = TAG-CT GACCGC--GCT
S4 = TAC----TCAC--GACCGACAGCT
Q1 = TAAAAC S2 S4 S3

13. Align Sequence S1 S2 S4 S3 S1 = -AGGCTATCACCTGACCTCCA-AA
S2 = TAG-CTATCAC--GACCGC--GCA
S3 = TAG-CT GACCGC--GCT
S4 = TAC----TCAC--GACCGACAGCT
Q1 = T-A--AAAC S2 S4 S3

14. Place Sequence S1 S2 S4 S3 S1 = -AGGCTATCACCTGACCTCCA-AA
S2 = TAG-CTATCAC--GACCGC--GCA
S3 = TAG-CT GACCGC--GCT
S4 = TAC----TCAC--GACCGACAGCT
Q1 = T-A--AAAC S2 S4 Q1 S3

15. Phylogenetic Placement
Align each query sequence to backbone alignment:
HMMALIGN (Eddy, Bioinformatics 1998)
PaPaRa (Berger and Stamatakis, Bioinformatics 2011)
Place each query sequence into backbone tree, using extended alignment:
pplacer (Matsen et al., BMC Bioinformatics 2010)
EPA (Berger et al., Systematic Biology 2011)
The two basic steps in phylogenetic placement is similar to the two phase methods we typically use, align and then place the fragment.
The differences are we align each fragment independently into the backbone tree. We then take each extended alignment and place each fragment independently into the backbone tree.
Two methods to align the query sequences are HMMALIGN, which estimates a hidden markov model on backbone alignment, then aligns query sequence. PaPaRa, which estimates ancestral parsimony state vectors for each in the reference tree and aligns the query sequence against each state vector. It selects the best scoring alignment.
Two methods that place the query sequence try to optimize the same criterion: find the ML placement given the extended alignment

16. Phylogenetic Placement
Align each query sequence to backbone alignment:
HMMALIGN (Eddy, Bioinformatics 1998)
PaPaRa (Berger and Stamatakis, Bioinformatics 2011)
Place each query sequence into backbone tree, using extended alignment:
pplacer (Matsen et al., BMC Bioinformatics 2010)
EPA (Berger et al., Systematic Biology 2011)
The two basic steps in phylogenetic placement is similar to the two phase methods we typically use, align and then place the fragment.
The differences are we align each fragment independently into the backbone tree. We then take each extended alignment and place each fragment independently into the backbone tree.
Two methods to align the query sequences are HMMALIGN, which estimates a hidden markov model on backbone alignment, then aligns query sequence. PaPaRa, which estimates ancestral parsimony state vectors for each in the reference tree and aligns the query sequence against each state vector. It selects the best scoring alignment.
Two methods that place the query sequence try to optimize the same criterion: find the ML placement given the extended alignment

17. HMMER and PaPaRa results
Backbone size: 500
5000 fragments
20 replicates
0.0
Increasing rate evolution

18. Old approach using single HMM
Leaves

19. Old approach using single HMM

20. Old approach using single HMM
Large evolutionary diameter

21. New approach

22. New approach

23. New approach
Smaller
evolutionary diameter

24. New approach
HMM 1
HMM 2

25. New approach HMM 1 HMM 2 HMM 3 HMM 4
the bit score doesn’t depend on the size of the sequence database, only on the profile HMM and the target sequence
HMM 3
HMM 4

26. SEPP (10% rule) Simulated Results
Backbone size: 500
5000 fragments
20 replicates
0.0
0.0
Increasing rate evolution

27. Using SEPP Known Full length Sequences, Fragmentary Unknown Reads:27
Using SEPP
Known Full length Sequences,
and an alignment and a tree
Fragmentary Unknown Reads:
( bp long)
(500-10,000 bp long)
ACCG
CGAG
CGG
GGCT
TAGA
GGGGG
TCGAG
GGCG
GGG .
ACCT
ML placement
40%
The two basic steps in phylogenetic placement is similar to the two phase methods we typically use, align and then place the fragment.
The differences are we align each fragment independently into the backbone tree. We then take each extended alignment and place each fragment independently into the backbone tree.
Two methods to align the query sequences are HMMALIGN, which estimates a hidden markov model profile for the backbone alignment, and then aligns the query sequence to the model, and PaPaRa, which estimates ancestral parsimony state vectors for each in the reference tree and aligns the query sequence against each state vector. It selects the best scoring alignment.
Two methods that place the query sequence try to optimize the same criterion: find the ML placement given the extended alignment
TAGA...CTT
(species3)
AGC...ACA
(species4)
ACT..TAGA..A
(species5)
AGG...GCAT
(species1)
TAGC...CCA
(species2)

28. Taxonomic Identification using Phylogenetic Placement28
Taxonomic Identification using Phylogenetic Placement
Adding Uncertainty
Known Full length Sequences,
and an alignment and a tree
Fragmentary Unknown Reads:
( bp long)
(500-10,000 bp long)
ACCG
CGAG
CGG
GGCT
TAGA
GGGGG
TCGAG
GGCG
GGG .
ACCT
2nd highest likelihood
placement 38%
ML placement
40%
Motivate the placement with numerical numbers 95%
TAGA...CTT
(species3)
AGC...ACA
(species4)
ACT..TAGA..A
(species5)
AGG...GCAT
(species1)
TAGC...CCA
(species2)

29. Taxonomic Identification using Phylogenetic Placement29
Taxonomic Identification using Phylogenetic Placement
Adding Uncertainty
Known Full length Sequences,
and an alignment and a tree
Fragmentary Unknown Reads:
( bp long)
(500-10,000 bp long)
ACCG
CGAG
CGG
GGCT
TAGA
GGGGG
TCGAG
GGCG
GGG .
ACCT
2nd highest likelihood
placement 38%
ML placement
40%
Motivate the placement with numerical numbers 95%
TAGA...CTT
(species3)
AGC...ACA
(species4)
ACT..TAGA..A
(species5)
AGG...GCAT
(species1)
TAGC...CCA
(species2)

30. TIPP
Nguyen et al. Bioinformatics 2014

31. TIPP for Taxonomic Profiling
Motivate 31
TIPP for Taxonomic Profiling
Marker-based abundance profiler
Uses a collection of single copy housekeeping genes
Only fragments binned to marker genes classified
Profiling algorithm
Bins fragments to marker genes
Classify fragments binned to each marker
Pool all classified reads
Estimate abundance profile on pooled reads

32. Taxonomic Profiling Experimental Design
5/14/14
Taxonomic Profiling Experimental Design
Datasets
Easy conditions (low error rates, known genomes)
Hard conditions (novel genomes, high error rates)
Methods
Marker-based – TIPP, Metaphyler, mOTU, Metaphlan
Genome-based – NBC, PhymmBL
Measured distance to true profile as error metric

33. “Easy” genome datasets

34. High indel datasets containing known genomes
“Hard” genome datasets
High indel datasets containing known genomes
Note: NBC, MetaPhlAn, and Metaphyler cannot classify any sequences from at
least of the high indel long sequence datasets. mOTU terminates with an error
message on all the high indel datasets.

35. “Novel” genome datasets
Note: mOTU terminates with an error message on the long fragment datasets and high indel datasets.

36. Summary
TIPP: marker-based taxonomic identification and classification method through phylogenetic placement
Very robust to sequencing errors and novel genomes
Results in overall more accurate profiles
Accurate profiles can be obtained by classifying reads from the marker genes

37. Acknowledgements Siavash Mirarab Mihai Pop Tandy Warnow Bo Liu
Supported by
NSF DEB
University of Alberta

38. SEPP/TIPP/UPP SEPP/UPP/TIPP site: https://github.com/smirarab/sepp/
Instructions for installing UPP:
Instructions for installing TIPP:
References:
1) N. Nguyen, S. Mirarab, K. Kumar, and T. Warnow. Ultra-large alignments using phylogeny-aware profiles, Proceedings of Research in Computational Biology (RECOMB) 2015 and to appear in Genome Biology 2015.
1) N. Nguyen, S. Mirarab, B. Liu, M. Pop, and T. Warnow. TIPP:Taxonomic Identification and
Phylogenetic Profiling. Bioinformatics, 2014, 30 (24):
2) Mirarab, S., N. Nguyen, and T. Warnow, SEPP: SATe-Enabled Phylogenetic Placement. Pacific Symposium on Biocomputing.

39. Phylogenetic Placement
Align each query sequence to backbone alignment:
HMMALIGN (Eddy, Bioinformatics 1998)
PaPaRa (Berger and Stamatakis, Bioinformatics 2011)
Place each query sequence into backbone tree, using extended alignment:
pplacer (Matsen et al., BMC Bioinformatics 2010)
EPA (Berger et al., Systematic Biology 2011)
The two basic steps in phylogenetic placement is similar to the two phase methods we typically use, align and then place the fragment.
The differences are we align each fragment independently into the backbone tree. We then take each extended alignment and place each fragment independently into the backbone tree.
Two methods to align the query sequences are HMMALIGN, which estimates a hidden markov model on backbone alignment, then aligns query sequence. PaPaRa, which estimates ancestral parsimony state vectors for each in the reference tree and aligns the query sequence against each state vector. It selects the best scoring alignment.
Two methods that place the query sequence try to optimize the same criterion: find the ML placement given the extended alignment

40. 16S Identification A A A A B B
16S gene

41. 16S Identification True Abundance A: 67% B: 33% A A
Estimated Abundance
A: 50% B: 50% A A B B
16S gene

42. Single copy gene True Abundance A: 67% B: 33% A A Estimated Abundance