Moore Notes 2 12 14

Subgroup to discuss SIMAP

• Participants: Katie, Stephen, Tom, Patrick

• What are our specific goals?
  • Stephen: gene catalog of gut microbiome genes (from shotgun metagenomes) and map to annotations
    • Tom: what about novel genes only in metagenomes?
    • Stephen: could do some assembly from shotgun metagenomes
    • Tom: need to avoid chimeras (see Bork paper)
    • Patrick: could SIMAP be used to identify chimeras?
    • Stephen: value is one mapping of reads from a sample to gene catalog and then to many gene families/annotations
    • Stephen: main challenge is partial length sequences (could focus on full-length gene predictions, though these have error)
  • Katie: same for marine gene catalog
  • Tom: same for other phylogenetically diverse communities (i.e., divergent from databases)
  • Updating genome-based gene family resources (e.g., Sfams)
    • Tom: value of this is marginal potentially, though HMMs are useful for new genome annotation
    • Probably greatest value is for environments that are not well studied, do not have lots of metagenomes
  • Patrick: expanding smaller databases such as BioCyc
    • Will follow up with Peter Karp on Feb 28

• Can these goals be fulfilled using existing protein family resources (eggNOG, figfams, BioCyc, KEGG, etc.)?
  • Not metagenomic gene catalogs
  • Versus updating Sfams
    • Benefit of Sfams is more diversity
    • Cost is lots of effort

• If we were interested in a larger scope project
  • Might be interesting to integrate other kinds of information into the protein family network.
    • e.g., protein family co-occurence (within genomes, operons, environments, samples, etc.).
    • This could give us a hint as to the function of novel protein families (if they co-occur with a known family)
    • Could also help us discover completely novel gene clusters, protein complexes, metabolic pathways, etc.
  • Tom is writing a proposal about environmental co-occurence
    • Depends on having sufficient number of metagenomes
  • Genome co-occurence and operon co-occurence would be very useful for getting at function
    • Tom: also helps with genome annotation
    • Tom: different ways of defining function are useful

• If we were to build families which sequences would we be interested in?
  • Specific to an environment?
  • To a set of taxa?
  • Do they need to be full length?
  • Include metagenomic data?
    • We would likely need to come up with some new strategies to deal with clustering partial length sequences into families.

• Would this be a limited-scope project mainly targeted for a specific application?
  • Example: identification of novel protein families in the gut? or surface ocean waters?
  • Versus larger project designed to provide a new kind of resource to the community
  • Focus on what we need to do our science
  • Start with a specific application for our lab then see if a resource is a good idea

• What SIMAP searches might we need
  • Genes vs. genes within catalog
  • Genes in catalog vs. other gene sets
    • For these two, SIMAP may be useful
    • Stephen will contact them
  • Reads vs. gene catalog
    • Probably not SIMAP, but bow tie or something similar