Subsystem Approach to Genome Annotation

1
Subsystem Approach to Genome Annotation

• National Microbial Pathogen Data Resource
• www.nmpdr.org
• Claudia Reich
• NCSA, University of Illinois, Urbana

2
Complete Microbial Genomes

• 464 complete microbial genomes in NCBI as of
  3-1-07
• 691 microbial genomes in progress as of 3-1-07

3
Making Sense of Genome Data

• Locate Genes identify ORFs automatically
• GeneMark
• NCBIs ORF Finder
• Glimmer
• Critica
• Assign Function by sequence similarity to
  experimentally characterized proteins
• BLAST family of sequence comparison tools

4
Problems with Assignments by Similarity

• When ORF is a member of a protein family
• Paralogous genes
• ORFs encoding similar proteins acting on
  different substrates
• Assignments can be transitive, and many times
  removed from experimental data

5
Other Factors Can Aid in Function Assignments

• Molecular phylogeny
• Paralogous and orthologous families
• Conserved gene neighborhood
• Metabolic context
• Bidirectional best hit matches across multiple
  genomes

6
Incorporating Information Other Than Similarity

• KEGG manually curated pathway and metabolic maps
• GO vocabularies that describe ORFs as associated
  with
• biological processes
• cellular components
• molecular function
• MetaCyc experimentally elucidated metabolic
  pathways

7
What is Needed

• A system that
• integrates all the above concepts
• organizes genomic data in structured idioms
• allows high-throughput annotation of newly
  sequenced genomes
• resolves discrepancies in different annotation
  tools
• informs experimental research

8
Enter the SEED

• Database and annotation environment
• Underlies, and accessible through, NMPDR
  (www.nmpdr.org)
• Expert annotation via subsystems building
• Provides the most accurate genome annotations
  available
• Argonne National Lab, University of Chicago,
  UIUC, FIG

9
What is a Subsystem?

• Any organizing biological principle
• metabolic pathway
• amino acid biosynthesis, nitrogen fixation,
  glycolysis
• complex structure
• ribosome, flagellum
• set of defining features
• virulome, pathogenicity islands
• functional concept
• bacterial sigma factors, DNA binding proteins

10
Subsystems are

• Sets of functional roles, which are functions, or
  abstractions of functions (such as an EC number),
  that together implement a specific biological
  process or concept
• Created manually by expert curators
• Experts annotate single subsystems over the
  complete collection of genomes, thus contributing
  and sharing their expertise with the scientific
  community

11
How Subsystems are Built

• Create a subsystem for the biological concept,
  and define the functional roles
• In one (or a few) key organisms that include the
  subsystem, find the genes and assign meaningful
  functional names
• Project the annotations to orthologous genes
• Expand to more genomes, creating a Populated
  Subsystem

12
Populated Subsystems

• Are Spreadsheets where
• Columns functional roles
• Rows specific genomes
• Cells genes in the organism that implement the
  functional role

13
How to Access Subsystems

• From Home page (left navigation bar) Subsystem
  Summaries select organism
• From Organism pages
• From Subsystem Search
• From protein pages to specific subsystems

14
Subsystem Pages in NMPDR

• Table of Functional Roles
• Subsystem diagram (if appropriate)
• Populated subsystem spreadsheet
• Customizable spreadsheet viewing options
• Functional variants and subsets of roles
• Curators notes

15
Benefits of Subsystems

• More accurate annotations
• Annotation of protein families
• Analysis of sets of functionally related proteins
• Less error-prone to automatic projections to
  novel genomes

16
Subsystems Reveal Interesting

• Pathway variants
• Are they clustered by phylogeny?
• Delta subunit of RNA polymerase only Bacillales
• Are they clustered by functional niche?
• Horizontal gene transfer?
• Fused genes
• ? and ? subunit of RNA polymerase fused in
  Helicobacter
• Fissioned genes
• ? subunit of RNA polymerase is fissioned in
  Cyanobacteria

17
Subsystems Reveal Interesting

• Duplicate assignments
• More than one gene for one functional role?
• Alpha subunit of RNA polymerase in Magnetococcus
  and Francisella
• Same sequenced region in more than one contig in
  partially assembled genomes?
• Frameshifts or other sequencing errors?
• Annotation errors?

18
Subsystems Reveal Interesting

• Missing genes
• Is the function essential?
• Is the function conserved?
• Does the missing gene cluster with homologs in
  other organisms?
• Is the function performed by a newly recruited
  gene?
• Has a gene been acquired by horizontal gene
  transfer and now performs that function?

19
Synthesis of Selenocysteinyl-tRNA

• Two known pathway variants
• One step in Bacteria
• SelA is annotated
• Two steps in Archaea and Eucarya
• PSTK was missing until very recently

20
Explore Selenocysteine Usage

• Start by searching for gene name, selA, in an
  organism known to use Sec, E. coli K12
• Start from subsystem tree expand category of
  "Protein metabolism," expand subcategory of
  "Selenoproteins"
• Open "Selenocysteine metabolism" subsystem from
  protein page or SS tree
• Genomes arranged phylogenetically
• Roles defined on mouse-over
• What genes are missing in which organisms?
• Are there Sec metabolism genes present in any
  organisms that do not have proteins that need
  Sec?
• Are there organisms known to need Sec for certain
  proteins, but that do not have a complete Sec
  biosynthesis pathway?
• Why is there a hypothetical protein included in
  this subsystem?