Moore Notes 1 15 10

OTU group call: JG, SK, KP, TS, SR, JL

• Mitch Sogin manuscript
  • JG: looks at three attributes that affect OTU estimation
    • PCR/seq error,
    • multiple seq alignment (overestimates) vs. pairwise,
    • linkage method in clustering (furthest neighbor overestimates), prefer average linkage
  • Analyze a bunch of data sets where they know the truth (experimental mixtures, or single species)
  • TS: how does it affect us?
    • pairwise alignment isn't an option for us.
    • user needs to quality control sequencing data prior to our pipeline (sequencing, but not PCR could be issue for metagenomics)

• Paper planning
  • What is novel in our work?
  • What will be cited? (Software? Paper?)
  • Do we hit a novel part of the biosphere?
  • Data comparison idea: PCR vs. metagenomic 16S from GOS
  • Chao or similar for upper bound and estimate how much underestimation going on out there in different data sets
  • May need simulations
    • how to compare distance matrices, beyond path difference (need to account for the fact that there are metric distances)

• To do list
  • Trees OK? Simulations (Sam with help from Tom & Steve on distance matrix evaluations, ecological evaluations)
  • Package code (Tom)
  • Code performance improvements (Tom)
  • Data analyses: (Tom and Steve will do some planning, maybe Josh and James can turn the crank once code is ready)
    • Re-analyze some other people's data sets?
    • PCR vs. metagenomic data analysis in GOS
    • Analyze new data sets (e.g. Quince et al data sets, what's new on CAMERA?, diversity map using all data sets we can find if enough)
    • Is diversity being underestimated? How much?