The idea is broadly:
- use ontology to create relationships (@balhoff help to extract) using phenex
- use relationships to inform ML
- recreate character states
- fill in a character matrix
- cluster phenotypes and genotypes
This idea stemmed from talks with @pmabee, @uyedaj, and Anuj Karpatne, and from reading Pasan Fernando's paper (methods for clustering analyses using ontologies and highlighting candidate genes) and @sergeitarasov (methods for reconstructing character states).
We could also compare the outputs of