Needs Analysis Workshop/Correlation Breakout

From phenoscape

Arhat Abzhanov, Todd Vision, Elizabeth Jockusch, Hopi Hoekstra, Jim Balhoff

Notes

Whiteboard

AA - Important to note difference in naming conventions between genetic terminology and diversity terminology

TV - Ontologies are meant to solve this problem - can map synonym terms to each other

EJ - many natural phenotypes are much more subtle than genetic mutants

TV - that is an advantage of relationships in ontology - can tell that phenotypes are related to the same morphology

AA - only some phenotypes will overlap - the types of variation are different

TV - but at least both may be annotated as affecting the same part of morphology

AA - user may want to be able to see traits at various levels within ontology - see everything within the tail or just one vertebra


TV - Covariance matrix can provide table of co-occurring changes in entities on tree

EJ - there are many complicated ways for doing character reconstruction - user will probably want to get data and run themselves

HH - especially in the case of parsimony, can't always treat transition probabilities as equal

HH & EJ - maybe provide some default methods, such as parsimony, maybe allowing specifying of transition models

EJ - co-ocurrrence table will be useful for eyeballing interesting groupings

EJ - also want to see which traits are related to the same genes

HH - this should be either location of expression in both entities, or same type of effect in both entities

On the branches where trait 1 changes, how many also have change in trait 2? Of the genes affecting trait 1, how many also affect trait 2?

  • View these in two matrices - are there correlations in these two matrices?

Should we allow the user to specify the entity-quality combinations they want in their matrix?

In mayflies, does having a particular gill morphology, increase diversity in habitat? Is gregariousness more likely to evolve in aposematically colored lineages?

I see these two traits covarying - how does the system tell you that?

  • table output (matrix) lets you visually scan for for co-occurrence
  • click on cell, bring up tree with covarying branches highlighted
    • or click on multiple cells and limit tree to those
  • further information on how the character changed...

2 approaches - interested in investigating a particular branch vs. interested in trait correlation across independent changes across tree

Gene can affect multiple tissues/structures

What genes affects all x structures? See results on anatomy view, separated into developmental series / developmental modules

  • See list of genes, click on gene, see mutant phenotypes, then see links to similar evolutionary phenotypes
  • See matrix of co-ocurring mutant phenotypes in two entities/ developmental regions

See colored indication on anatomy view of co-occurrence showing shared genetic bases

Find minimum number of genetic changes which can account for a particular compound evolutionary change (multiple phenotypes)