PCGR integrates a variety of different datasets, that typically use different notations or ontologies when it comes to diseases or phenotypes. We have cross-referenced a variety of ontologies (DiseaseOntology, EFO, OncoTree, UMLS) in use in the oncology domain, and organized all the various tumor subtypes into 30 different main tumor sites (see also https://github.com/sigven/oncoPhenoMap).
Below you can browse all the various phenotype terms that are
associated with each of the primary sites. This as a guide to choose the
primary tumor type when running PCGR for a given sample (option
--tumor_site).