Closed tttor closed 6 years ago
This means that the feature vector for learning should include binary string [0,1,...] that encode plant composition, similar to ijahv1.
This implies we have to run 2 kinds of predictors: the first is with binary string features, and the second is with compound-protein-pharmacological level features
plant -> someCompound -> someProtein -> Efficacy/DiseaseClassification
This depends on #107
Superceded by #183
This refers to the old data used by Farit, Aries, ... for Ijahv1. This is also for compensating the fact that plant-compound data are scarce.