Paramvir S Dehal, Jeffrey L. Boore

1
A phylogenomic gene cluster resource the
Phylogenetically Inferred Groups (PhIGs) database

• Paramvir S Dehal, Jeffrey L. Boore

Presented By Jared Carter
2
Principal Objectives

• Define and add context to the growing need for
  better methodology for high-throughput sorting of
  genes into Orthologous families.
• Outline the steps involved in the computational
  framework for the identification of these sets.
• Examination of model output data
• Analysis of overall model effectiveness.

3
Introduction

• Increasingly large whole genome projects require
  further methodologies to increase the overall
  contextual value of the data set.
• Delving into the evolutionary history of each
  gene leads to a more robust understanding of its
  function.
• Gives definition to the evolution of the genome
  and specie.
• Answers questions regarding the functional and
  biochemical processes of the genome.
• Previously homologs identified by pairwise
  comparisons has been used, with many drawbacks.
• Incorrect assignments caused by gene-duplication
  events
• Accelerated rates of AA substitution
• Domain shuffling

4
Methodology Contextual Background

• Uses known evolutionary relationships to
  construct gene clusters of queried genomes.
• Analyzes each cluster for evolutionary
  relationships between the queried genes.
• Goal is to reconstruct the evolutionary history
  of each family.
• Allows for whole genome analysis with emphasis on
  the evolutionary background of each gene.
• Can discriminate and classify numerous additional
  types of evolutionary events (gene duplications,
  AA substitution rates, and gene loss)

5
Methodology Overview

• Creates and populates a relational database with
  all known annotations for each taxon.
• All sequence data and annotations available are
  included
• Data such as previous sequence alignments and
  trees are also included
• General process involves 5 steps
• Step 1 All against All BLASTp
• Step 2 Global alignment and distance calculation
• Step 3 Hierarchal Clustering
• Step 4 Multiple Sequence Alignment (MSA)
• Step 5 Gene Tree Construction

6
Methodology Workflow Diagram
Figure 1 Work flow diagram illustrating analysis
pipeline for processing select gene models. 1
7
BLASTp and Global Alignment

• All against All BLASTp generates local alignments
  which then must be processed into a global
  alignment.
• ClustalW is used to align each protein pair.
• These resulting alignments are stored in the PhIG
  database
• Distances are calculated using the JTT matrix
  used in the ProDist program (PHYLIP)
• Used later in the clustering process in
  conjunction with the gap-free alignment lengths

Figure 2 BLASTp query output excerpt. 2
8
Gene Cluster Analysis

• Process takes in the known evolutionary
  relationships of the selected organisms as well
  as the all pairwise protein distances
• Uses an iterative approach, starting at the base
  of the best known evolutionary tree at the common
  ancestral gene, and then extends upwards
• For each bifurcating node two new clades, A and
  B, are created, with the remaining taxa being
  added to an out-group.
• Genes from clade A are more similar to each other
  than in clade B.
• Similarly genes from A and B are more closely
  related than those in the out-group.
• This is accomplished by comparing seeds (pairs of
  sequences) and assigning a match quality.
• Clusters grow by determining the shortest
  distance from A and B.
• If Proteins have a shorter distance than A-B it
  is added to the cluster
• This is repeated until all proteins have been
  clustered

9
Protein Distance Map and Cluster Tree
Figure 4 Species tree generated by cluster
analysis.1
Figure 3 Protein Distance map of a pairwise
alignment.1
10
MSA and Phylogenetic Tree Extrapolation

• An MSA is performed on each cluster, using
  ClustalW.
• Alignments are trimmed to remove gaps, or removed
  if fewer than 100 AA are aligned successfully.
• Trees are generated using a precise method
• The quartet puzzling maximum likelihood method,
  implemented by TREE-PUZZLE using JTT
• The generated tree is assisted by the known
  evolutionary relationships, further defining the
  phylogeny.
• MSAs are also used to find Hidden Markov Models,
  which give better structural representation to
  sparsely sampled genomes.

11
Phylogenetic Tree
Figure 5 Phylogenetic Tree incorporating the
cluster tree and known phylogenetic data.
Illustrates numerous duplication events as well
as the proportionality of AA substitutions
(represented by varying branch lengths). 1
12
Gene View, Annotation and Synteny Maps

• Gene View
• All known annotations to a gene can be displayed,
  as well as its location on the chromosome of all
  selected species.
• Querying
• The database can be searched for exact and
  relative sequences, or text matches to search
  fields.
• Several specialized search tools exist, such as
  HMMER, which searches directly against Hidden
  Markov Models.
• Synteny Maps
• Produce a set of one-on orthologs plotted in
  their relative locations across multiple genomes.
  
• Performed by selected a reference sequence span
  of one species, and the species to be queried.

13
Synteny Map
Figure 6 Synteny map of selected sequence with
rectangles representing genes, to the left and
right orthologs. Note Black lines indicated
similar transcriptional orientation, red lines
indicate inversion.1
14
Model Effectiveness Overview

• Very efficient at analyzing evolutionary patterns
  within whole genomes.
• Assists in transferring annotations
• Identifies sources of evolutionary pressure on
  genes
• Utilizes the applicable known evolutionary
  history of the genomes studied
• Creates clusters as descendants from one
  ancestral gene
• Combines phylogenetic data with functional
  annotation, gene structure and genomic position.
• Multiple, varied applications
• Organismal Phylogeny construction
• Genome evolution via gene duplications
• Gene structure evolution via gain/loss of exons,
  introns and domains.
• Identification of Gene family expansions and
  losses
• Genome evolution as a whole.

15
Conclusions

• With continuing development this highly modular
  tool which is well founded in current analytical
  molecular methodology will prove to be quite
  useful in adding historical context to genome
  data.
• Understanding the history of a gene allows for a
  better realization of that genes function in the
  organism by relating it to other, similar
  organisms which have also conserved its function.

16
Resources

• 1 Dehal, Paramvir S., Boore, Jeffrey L., A
  phylogenomic gene cluster resource the
  Phylogenetically Inferred Groups (PhIGs)
  database, BMC Bioinformatics, 2006, 7201
• 2 NCBI, BLASTp protein-protein query,
  http//www.ncbi.nlm.nih.gov/BLAST/Blast.cgi