cBioPortal
Category: Cancer study package
cBioPortal is a platform and study format for cancer genomics data. Convert-Pheno reads study metadata, patient and sample clinical tables, and case lists from a study package. This profile is implemented and regression-tested; timeline and molecular files are not converted.
What You Needβ
Supply either an unpacked study directory or a .zip archive. The package
must contain:
meta_study.txt- one sample-clinical metadata file and its referenced data table
- optionally, a patient-clinical table and
case_lists/
convert-pheno -icbioportal study/ \
-obff --entities individuals biosamples datasets cohorts \
--out-dir bff_out/
PXF and OMOP-CDM output use the same input. See the cBioPortal recipes.
What Is Convertedβ
| cBioPortal content | BFF content |
|---|---|
| study metadata | datasets |
| patient rows | individuals |
| sample rows and patient links | biosamples |
| case lists | cohorts |
ONCOTREE_CODE | OncoTree: disease and histological-diagnosis terms |
Identifiers and links are checked before mapping. Duplicate identifiers or references to unknown patients and samples stop the conversion.
SAMPLE_TYPE values such as Primary are cBioPortal source categories, not
necessarily ontology concepts. They remain in source provenance, while
sampleOriginType stays Not Available unless a project mapping supplies a
reviewed ontology term.
Use --no-source-info to omit the copied patient, sample, and study metadata.
Structural links and mapped semantics remain.
Optional Project Mappingβ
A Mapping V2 file with source.profile: cbioportal can map additional patient
beacon.biosamples evaluates sample attributes. Source PATIENT_ID and
SAMPLE_ID values cannot be rewritten because sample links and case lists
depend on them.
See Mapping Files and cBioPortal to BFF for the detailed contract.
The package is processed in memory. ZIP archives are read directly. Timeline and molecular files are not converted, and this filesystem-package route is not exposed through the HTTP(s) API.