Moore Notes 2 27 13

Group Call

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

• In person meeting
  • Samantha can't make June 19
  • Move to June 17
  • San Francisco

• Tom: Needs Dongying's help getting Interpro IDs for IMG genomes

• Josh: Beta diversity niche mapping
  • Community classification
    • What clustering algorithm? UPGMA seems slightly better than other hierarchical clustering algorithms
    • How many clusters?
  • Not finding distinct clusters - why?
    • Check if find distinct clusters with marine macroorgansim data
    • Bray-Curtis (abundance based) might be flat due to a couple OTUs that dominate - try presence/absence
    • Look at actual Bray-Curtis matrix (ordered by clusters) to confirm no visual patterns
    • Repeat for individual phyla
  • Survey gap analysis
    • Use model to estimate how dissimilar an un-surveyed grid cell is from surveyed locations
    • Points to locations likely to have novel communities
    • Tom: How to validate? Josh: Independent data set, cross-validation; Katie: Collect data
    • Tom: Do they correspond to low certainty locations? Josh: No, not precision. Difference.
    • Sounds cool...

• Dongying: mutual information based clustering approach
  • Will give update in two weeks

• New iSEEM2 postdoc starting in April (Sarah Hird)

• New bioinformatics staff member at Gladstone (Tim Laurant)

• MMI meeting next week
  • Katie and Russell will be there