Moore Notes 7 10 13

From OpenWetWare
Jump to navigationJump to search

Group Call

  • Participants: Katie, Stephen, Tom, Josh, Ladan, Dongying
  • Dongying: wants to do metagenomic simulations for TreeOTU validation
    • Wants to simulate a community
    • Reads need to be >200bp
    • Stephen has simulation pipeline using Grinder with communities pre-defined
    • Also need to calculate 16S copy number (try Steve Kembel's tree based inference of this)
  • Josh: talked to Jack Gilbert about samples
    • Has a call scheduled with him next week (Monday 10am)
    • Jack will put people with samples on call
    • Josh will follow up with Jonathan
  • ncRNA gene families
    • Could scan English Channel unmapped reads in similar manner to HOT/ALOHA ncRNA paper:
    • Stephen: could be hard to build de novo families from short reads
      • Try assembly
      • Tom: might work because English Channel and HOT/ALOHA are 454 length reads
    • How hard is it to build a model?
      • Structure based alignment is challenging
      • SCFGs are slow
  • Another sifting step would be to run them against vFams for viruses
  • Stephen: been playing with metagenomic gene finders as an alternative to 6-frame translation
    • Reduce the volume of data to search in MRC six-fold
    • 5% reduction in sensitivity
    • Not clear yet if the filtered ones are biased or effect relative abundance estimates
    • Used fragGeneScan