Skip to main content
Fig. 2 | Journal of Biomedical Semantics

Fig. 2

From: Predicting instances of pathway ontology classes for pathway integration

Fig. 2

Bootstrapping procedure. The initial training data is derived from existing PW mappings and UMLS mappings between MeSH and GO. A simple logistic regression model is trained on this data and used to bootstrap training samples from Reactome. The best matches between Reactome pathways and PW classes are added to the training data set over 10 iterations to generate a final training data set

Back to article page