Moore Notes 12 8 14

Group Call

• Participants: Jonathan, Tom, Josh, Ladan, Sarah, Guillaume, Dongying, Patrick, Stacia, Stephen

• Patrick: Gene functions necessary for life in an environment or variable across samples
  • slides
  • Application: healthy human gut (lots of data)
    • JE: stool isn't really a single niche/environment
  • Taxonomic variability and functional stability previously noted
  • Model accounts for mean overall and per study
    • Mean correlated with variance (as expected for count data)
    • Different definitions of "healthy" across studies
    • Test statistic is the residual from the model (unexplained variance)
  • Two possible null hypotheses: high variance or high variance given mean
    • Null distribution by simulation from fitted model
    • Significantly variable and invariable gene families
    • JE: also show genes with significant study effects
  • Variable gene families are enriched for transporters, PTS systems, and nitrate metabolism
    • Variability and abundance correlation with taxonomic breadth of gene family
    • Abundance of these are correlated with average genome size
    • Ian Paulson, Milton Saier papers on transporter copy number and genome size
      • Broader specificity of transporters in smaller genomes
      • Subfamilies necessary to see differences, e.g., flux versus affinity
  • Invariable gene families are enriched for central metabolism, ribosome, vitamin biosynthesis, non-mevalonate isoprenoid biosynthesis (cell wall), exon junction complex (why?)
    • Stephen: Probably not host contamination (see plot)
  • Phylogenetic logistic regression to look for evidence of selection
    • Tests for association of gut habitat with presence/absence of gene family, across genomes
    • Gut depleted gene families are associated with invariability
    • JE: Risky to annotate taxa by environment, Louzapone 2008 annotation is old
    • Kostas (GA tech) ecological correlates of genome properties, Jenna in Eisen lab did some analysis for correlates with LGT