Moore Notes 2 13 13

Group Call

• Participants: Katie, Tom, Stephen, Josh, Guillaume

• In-person meeting
  • Date: June 17 (Mon) with June 19 (Weds) as backup, one day meeting no overnight
  • Location: tentatively SF

• Guillaume will visit SF soon

• SFams automatic updates
  • Tim Laurant will work on the QC side of the update pipeline
    • Precision and recall, network assessment, etc for new families
    • Will stitch together existing software into a pipeline for computing these metrics
  • Apply to all families at each update
    • Big compute as db grows, might need to improve efficiency
    • But maybe not - can run in background
  • Stephen's additional QC metrics

• LAST vs. BLAST for classifying reads into families (MRC, phyloSift)
  • After partitioning reads by genomes, looks like BLAST is better for diverse genomes
  • LAST is a lot faster though
  • For phyloSift could double check long branches with large m parameter or fixed length seed or several m's
  • Similar to TribeMCL

• Next week: Katie teaching