Moore Notes 7 13 11

Group Call

• Protein db
  • Data deliverables
    • Updates (e.g., as new genomes are sequenced)
    • Could be an SQL dump with an archive of flat files
    • Flat files (e.g., on Biotorrents)
      • alignments
      • mappings of sequences to families, which were used to build HMM
      • organize into meaningful, subsets (e.g., all with high universality, representative sequences for fast searches)
    • User interface?
      • Try to get CAMERA to do this, as proposed originally
  • Morgan's observations about other databases, why are we unique?
  • Journal
    • PLoS ONE
    • NAR database issue - slower and the issue is getting huge
    • BMC Bioinformatics
    • Briefings in Bioinformatics
  • Do we need to build trees? Seems like a good idea (Morgan)
    • Method? PhyML, FastTree
    • Should be pretty fast if we only build trees with the representative sequences
    • How to measure PD of a family from the tree? Needs to be comparable across families
  • Analyses to do
    • Finish cleaning up precision/recall analyses (Guillaume & Tom)
    • Scanning new genomes - 2500 genomes now vs. 1900 used to make db (Dongying)
    • Scanning GOS metagenomic data set - test compute time (Guillaume & Tom)