Moore Notes 10 16 13

From OpenWetWare
Jump to: navigation, search

Group Call

  • Participants: Katie, Stephen, Guillaume, Sarah, Ladan, Tom, Josh
  • Discussion of Jack Gilbert's work on PICRUSTS and niche modeling
    • PICRUSTs/MAP vs. shotgun metagenomes
      • Their previous analysis of English Channel data found high correlation
      • They used an old version of PICRUSTs plus their own pipeline
    • Differentiating our projects
      • Different scientific questions (changes in ecosystem services)
      • They are processing EMP data sets, we would process MICROBIS
  • Jack joined our call
    • Paper on predicting carbon flux in English Channel has been hard to publish
      • Concern about over extrapolating or over-fitting models or over-interpreting features in the data vs. directly measuring ecosystem services
    • Same techniques on Gulf of Mexico data
    • What other things Jack wants to do with niche modeling in future vs. us?
      • Jack: metabolic prediction, stability issues with key biogeochemical cycles associated with human activity (spatial and temporal activity of pollution events) using MAP/PMRT
      • Josh: functional diversity, survey gap analysis, extinctions
    • Doesn't want us to reinvent the wheel with English Channel data
    • They are doing functional predictions from EMP data, not MICROBIS
      • Could compare EMP vs. MICROBIS maps in a paper about sampling issues, reliability of methods, etc.
    • EMP primers are not good at picking up SAR11 (Furman noticed this)
    • Tara Oceans coming out soon
      • Will see where they sampled, then use EMP or call to community to fill gaps (e.g., based on conclusions of EMP vs. MICROBIS maps)
    • Josh: Do you know any one sampling marine hosts? Whales, sponges
      • Jack has looked at boats, also microbes on plants (merlot vines)
    • American gut, mouse guts around N. America - correlated?
      • Jack talked about running MAP on American gut, but metadata is challenging
      • 5-6K people with data from phones and shoes - this would be good for niche modeling
    • Tibet soil project might be of interest to Nature
    • Gulf of Mexico raster data