Background & rationale. Prostate, breast and colorectal cancers often present more aggressively in people of African ancestry, yet genome-wide association studies (GWAS) for these populations remain severely under-powered. Leveraging the self-identified Black/African UK Biobank participants-together with 1000 Genomes African reference panels-we will address this gap and improve equity in cancer genetics and possible improved precision medicine.
Research questions.
1. Which single-nucleotide polymorphisms (SNPs), genes and biological pathways underlie these three cancers in African-ancestry individuals?
2. How do risk loci, fine-mapped causal variants and polygenic risk scores (PRS) compare with findings in European-ancestry cohorts?
3. Can machine-learning (ML) models that incorporate African-specific effect sizes outperform conventional European-derived PRS in predicting cancer risk?
Objectives.
* Perform stringent QC, ancestry clustering and fine-scale sub-ancestry inference.
* Conduct single-variant GWAS with linear mixed models (LMM & EDLMM) and gene-based tests with SKAT.
* Integrate expression-weighted association tests (MOKA) and perform pathway enrichment (GO, KEGG, Cancer Mine).
* Apply LD-aware fine-mapping with MAGMA to prioritise likely causal variants.
* Build and benchmark ML classifiers and PRS (p < 0.05 filters) within African participants and compare performance to PRS trained in Europeans.
* Release summary statistics and code to advance cancer genetics in under-represented populations.
Impact. Identifying African-specific genetic risk factors will (i) improve our understanding of cancer aetiology, (ii) provide a foundation for ancestry-tailored screening and prevention strategies, and (iii) reduce disparities in genomic medicine.
