Moore Notes 3 2 15

Group Call

• Participants: Katie, Tom, Dongying, Sarah, Stephen, Patrick, Guillaume, Josh

• Shotmap manuscript ready to submit
  • Reviewers / conflicts of interest
  • Corresponding author

• Dongying: slides
  • Phylogenetic profiling of gene families
    • Presence/absences indicates functional relationship
    • Correlation assumes independence of lineages, which is not true
    • Independent contrasts approach corrects for dependence but creates errors
    • Solution uses TreeOTU at different cutoffs and looks at correlation over range of phylogenetic levels
    • Analyzed ~8K genomes and 9K PfamA families
      • Different patterns of loss/gain across lineages and Pfams
      • Many Pfams come and go across phylogeny
      • Identified core genes with consistent presence/absence
      • SPICi clustering of Pfams using average distance at 3 semi-independent phylogenetic levels
  • Next steps:
    • Maximum likelihood ancestral state reconstruction (vs. Felsenstein 1985)
    • Other protein databases
    • Visualization
  • Katie: Gold standards for true and false positives
  • Tom: what is gained versus using overall correlation (i.e., not using tree)?
  • Patrick: Is this a special case of TIC (phylogenetic regression), e.g., with weights?

• Future calls
  • March 16: Tara Oceans
  • March 30: Patrick
  • April 13: TBD