Moore Notes 1 7 09

Group Conference Call (skype)

• UCD Cluster
  • send logins to JE to send to Mike Lewis
  • how should people do analyses (cluster, Eisen lab linux machine bobcat)

• Srijak is working on data set of protein families
  • alignments w/ metagenomics reads from depth series
  • need to build trees

• Annual report due in ~3 weeks
  • JG and SK will make a wiki page about what they need (e.g. rRNA alignment/trees for assessing diversity)
  • JE lab will make a page about what datasets are available
  • JGI database? Check in literature for who has already done this?
  • GOS, depth series (e.g. using infernal)

• Simulations
  • Martin's data is on bobcat.genomecenter.ucdavis.edu/AMPHORA website
  • metaSim, JGI simulation paper (datasets available, not clear exactly how done)
  • Make a new method
    • Idea: hold out some complete genomes as "complete" and fragment the rest to look like metag. reads
    • What do reads look like: how long (depends on seq technology)? how much overlap (depends on community structure)?
    • supermatrix methods depend a lot on how this is done
    • simulation process should imitate community structure (rel abundances), molecular methods
    • would we do some assembling? better to focus on raw reads
    • should look at a range of parameters

• Tree-building
  • additional people to talk to: Lior Pachter, Brian Moore
  • supermatrix
  • supertree
    • for combining across gene families? how to map taxa?
  • "traditional" tree building with all data (is this different from supermatrix? maybe just in its use of missing data)
  • Delong paper: phylogenetic analysis on a depth series

• Planning:
  • Jenna, Steve, Sam will start a wiki page on simulations
  • Sam will email Martin to see if he can chat
  • JG will start a data wish list page