Moore Notes 3 4 09

Group Conference Call

• Data sharing
  • genbeo (behind a firewall - need to sftp, a few terrabytes)
  • put a note on the wiki with readme info
  • include non-metagenomic datasets, maybe in a subdirectory

• CollectiveX or other conversation system
  • how to have huge discussions? Let's try it.

• MetaSim
  • problems with command line
  • how many species to include and how related (random?)
  • fasta file vs. alignments to AMPHORA profiles (folks want alignments)

• ComboDb vs. AMPHORA
  • different sets of sequences (e.g. for rpoB)
  • almost all sequences in ComboDb are in AMPHORA profiles
    • ask Martin where he got the sequences
    • would be nice if we can link between ComboDb and AMPHORA with a unique id

• tree building
  • pplacer discussion
  • computational time for ML tree estimation
    • on 8 core computer full ML with 700 seqs = 9 hours without bootstrapping
    • using reference seq tree as a guide tree speeds it up to 1-2 hours
    • 100s of sequences are OK, 1000s are not
  • Steve's results: only with reads vs. with reference sequences (in two ways)
    • reference tree helps to parcel out reads, without it the reads get lumped
    • could be a problem for clades where we have reads but no reference seqs

• possible visit of Josh to Eugene