Needs Analysis Workshop/Report-out Day1

From phenoscape
Revision as of 16:33, 19 September 2007 by Hilmar (talk | contribs)

Evo Group report-out

  • need to improve ontologies, e.g. for describing behavior
  • need to improve tools for analyzing images
  • covariance of entities, attributes
  • broader queries of entities within particular constraints, such as structures
  • need capacity to store media (e.g., images, movies)
  • need environmental, ecological, and geographical data

Devo Group report-out

  • identifying homologies among distantly related taxa
    • for example, do fishes have digits? E.g., if genes have similar effects on fins as on digit development.
  • patterns of covariation
    • for example, cusps on mammalian teeth
    • could be used to identify "character modules" that are well-connected
    • connections within modules should be more stable than between
  • for a branch hypothesized to be under positive selection (from e.g. a molecular analysis), identify candidate phenotypes that could have been selected for
    • map synapomorphies onto a phylogeny
    • identify phenotypes with highest rate of change on that lineage
    • needs phylogeny that is selectable and manipulable
  • use the database to point out holes in the knowledge

Discussion

  • enrichment analysis: given a list of genes, which phenotype terms are overrepresented among them
  • capturing representing variation seems to be a general problem and challenge, but critically needed
    • variation in regulatory (or morphogenic) gene expression and corresponding phenotypic changes (Hox gene expression example)
    • polyphenism: variability of phenotype that can distinguish species but is also dependent on environmental factors or variable levels of gene expression
    • do we need to capture frequencies for polymorphisms to represent variability
    • there are morphological differences even between inbred strains, for example for JAX mouse strains in cranial morphology
  • integrating with genomic resources:
    • linking out to genome databases seems mandatory
  • alerting features
    • for example, saving queries as RSS feeds