Moore Notes 4 16 14

From OpenWetWare
Jump to navigationJump to search

Shotmap group call

  • Participants: Katie, Tom, Patrick, Guillaume, Stacia, Stephen
  • In-person meeting May 8: discuss details next time
  • Patrick: Building trees for KEGG and Bio/MetaCyc families
    • Clustal-omega alignments
      • Check that it does full-length global alignment
      • MUSCLE good for low percent sequence identity, but Clustal-omega now doing the same or better
    • Tree building: RAxML vs. PhyML?
      • Tom: PhyML faster and only a bit more error
      • Could also try FastTree
      • No alignment masking?
    • Goal: estimate phylogenetic diversity of each family
      • Normalize by number of members (paralogs)?
      • Will be depressed in MetaCyc which only uses representative genomes
      • Restrict to phyla represented in microbiome? (trim tree after, build with all)
  • Shotmap paper progress
    • Stacia got figfams (release 12 version), classifying L4 with shotmap
      • Later (2012) release doesn't have fasta files or documentation
      • No longer funded, taken over by PATRIC
      • Stephen: Will we release formatted family dbs? No - too many issues releasing, hosting, and maintaining
      • Patrick helping her debug shotmap runs
      • Will compare to other protein dbs (with old classification thresholds)
    • Stephen/Tom: Effect of read length via simulated metagenomes
    • Tom: software updates
      • Implementing multi-threading tool to specify processors on a stand-alone server
      • How to speed up database above millions of rows per table? Working on db-free classification procedure
      • Also making a totally db-free version (for speed, storage)
      • Error in mySQL query when computing abundances
  • Next steps
    • Finish L4 shotmap analysis and comparison with other dbs
    • Finish threshold optimization
    • Reclassify all data sets
    • Address software loose ends
      • Install issues
      • Log verbosity
      • Documentation gaps (e.g., quick-start read me)
      • Small test example
  • Check out new version in 1.5 weeks to make sure we are all using the same version
  • Patrick: pull requests in github