Logo Lanfrica

Behordeun/afripharmagen-catalog

Domain:

healthcare

Record type:

dataset
Creator:
Beh
Host:
Curated pharmacogenomic data and reproducible analyses for sub-Saharan African populations # afripharmagen-catalog Curated pharmacogenomic data and reproducible analyses for sub-Saharan African populations. ## Contents ### data/ Core datasets: - `alleles/african_alleles.json` - Machine-readable catalog of 10 clinically actionable pharmacogenomic alleles across 6 genes, with African-specific frequency ranges, functional annotations, and evidence levels. - `samples/1000g_african_samples.tsv` - Population metadata for 661 samples from seven 1000 Genomes Project African populations (YRI, LWK, GWD, MSL, ESN, ACB, ASW). ### results/ Benchmark outputs: - `afripharmagen_benchmark.jsonl` - Star-allele calls from afripharmagen for all 661 samples x 10 genes (6,610 results). ### analyses/ Reproducible analysis modules: - `concordance_benchmark/` - Comparison of afripharmagen vs. PharmCAT v3.4.0 carrier detection - `population_frequencies/` - Per-population allele frequency estimation, gradient analysis, and Wilson score confidence intervals - `allele_catalog/` - Catalog schema documentation ## Key Findings PharmCAT v3.4.0 cannot call CYP2D6 from standard variant-only VCF output (0/661 African samples callable), a known VCF format limitation documented by the PharmCAT team. afripharmagen's reduced-position strategy identifies 243 CYP2D6\*17 and 134 CYP2D6\*29 carriers from the same input without reprocessing. For other genes, both tools agree (95-100%). Three alleles show inter-population frequency gradients exceeding 10 percentage points across the seven African populations: CYP3A5\*1 (19.3pp), CYP2D6\*17 (13.5pp), and CYP2D6\*29 (13.1pp). Per-population frequency data with 95% Wilson confidence intervals are available in `analyses/population_frequencies/results/`. ## Data Sources - Allele definitions derived from PharmVar v6.0.4 and published literature - Whole-genome sequence data from the 1000 Genomes Project high-coverage dataset (Byrska-Bishop et al., Cell, 2022) - PharmCAT v3.4.0 (github.com) ## Qu …

Licenses