Moore Notes 7 16 14

Group Call

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

• Eugene META symposium in August
  • Katie, Stephen, Tom will coordinate presentations

• Shotmap paper
  • Stephen working on simulations
  • Processed all Spanish MetaHIT samples with read-length abundance thresholds
  • L4 samples all annotated and analyzed for temporal patterns
    • Some Gilbert patterns recapitulated, some not
    • Rapsearch is using a lot of disk (for temp files)
    • Rapsearch uses a lot of memory if you don't employ multiple threads
    • Issues for local installations (60 million reads creates a 1TB footprint vs. KEGG, 20 million reads also 1TB vs. Sfams)
    • Reducing the number of hits might help, but need to think about impact on downstream analysis
  • Stephen evaluated different approaches to read translation and gene prediction
  • Planning for paper via email, then discuss on July 30 call

• Dongying's new project about phylogenetic analysis of eukaryotic taxa in metagenomes (PHYECO eukaryotic markers)
  • Slides: http://edhar.genomecenter.ucdavis.edu/~dwu/presentation_dir/euktest.pdf
  • Laura Katz has identified 34 eukaryotic marker genes
    • Did not use same criteria as Dongying did for bacteria
    • Did not have very many genomes
  • Could also look for subclasses, such as diatoms or fungi
  • Challenges
    • Organelle derived genes
    • Low coverage
    • More copy number variants
      • Duplications and multiple copies of mitochondrial genome
      • Evenness not so useful
      • Universality and monophyly are more important
    • Huge volume of data
      • May need to divide into subgroups
  • Stategies
    • Expand bacterial/archaeal markers
    • Use 34 markers
    • Use tree to separate out the non-eukaryotic hits
      • Necessary because there is no score threshold that separates eukaryote from non-eukaryote hits
      • Works for a ribosomal protein, but not a mitochondrial protein
    • Focus on assignments to subgroups (lower down in trees), where they are monophyletic (i.e., do not need to resolve deep branches now)
    • Use metagenomic data to refine marker set (good data on universality, beyond what can be done with genomes)

• Patrick: related question - Is there some reason half of the archaeal ribosome would be variable across metagenomes (like the eukaryotic ribosome) and the other half would not (like the bacterial ribosome)?