Develop a Pharmacogenomics widget with PharmGKB data

ireneisdoomed commented 9 months ago

We want to integrate Pharmacogenomics data into the Platform, with the aim to inform our users about those variants that have an impact in the response of a drug. This will enable a more personalized approach to drug discovery and development, taking into account how specific genetic backgrounds can influence treatment outcomes.

Our initial plans are to have a first version of this PGx data widget in December, with the intention of keep adding new sources as we become aware of them.

Background

We focused on using PharmGKB as the central repository of this type of information. PharmGKB is a project that curates knowledge about the relationships between drugs, diseases/conditions, and genes, including details on genetic variants that influence drug metabolism, efficacy, and adverse effects.

Their section that have this relationships between a variant in a gene and their change in the drug response is what they call “Clinical Annotations”. We have basically worked with this dataset, parsing it and converting it into a data model that is harmonised with the rest of our data.

Data stats

Some exploration on clinical_annotations.tsv:

4919 evidence. No type of identifier is used for Genes/Phenotypes/Drugs.
All evidence have variant/haplotype information.
- Almost all (4347) represent a genetic variant.
2592 target/disease associations.
3478 target/disease/drug associations.
653 distinct phenotypes.
852 distinct drugs.
1061 distinct genes.

Tasks

[ ] Solidify the data structure
[ ] Design the widget for the target and drug pages
[ ] TBD

prashantuniyal02 commented 9 months ago

The schema details can be found here: #3110

DSuveges commented 9 months ago

The pharmacogenomics data has arrived:

Total clinical annotations: 5073
Total evidence strings: 24641
        With CHEBI: 20451 (83.00%)
        With EFO phenotype: 8139 (33.03%)
        With functional consequence: 15633 (63.44%)
        With VEP gene: 15633 (63.44%)   
Total RS: 2794
        With >2 alleles: 31 (1.12%)

Data is here: gs://otar012-eva/pharmacogenomics/cttv012-2023-10-12_pgkb.json.gz

opentargets / issues