Meshach-Zm/African-Variant-Triage-Pipeline

Converts African-cohort VUS from statistical associations into multi-evidence functional hypotheses, ready for wet-lab prioritisation # African Variant Triage Pipeline > **VCF input → 3-Layer Evidence Convergence → Prioritised Wet-Lab Action List** A multi-layer computational pipeline that converts African-cohort variants of uncertain significance (VUS) from statistical associations into multi-evidence functional hypotheses, ready for wet-lab prioritisation. --- ## At a Glance | Input | Process | Output | | ---------------------------------------------------- | --------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------- | | Tab-separated VCF (variant_id, CHROM, POS, REF, ALT) | 3-layer evidence convergence: deep learning + empirical databases + population genetics | Ranked variant shortlist with convergence scores, evidence summaries, and specific wet-lab assay recommendations | --- ## Table of Contents - Overview - Why This Pipeline vs CADD / VEP - Pipeline Architecture - Project Structure - Installation - Running on Google Colab - Quick Start - Input Format - Usage - Output Files - Evidence Convergence Scoring - Validation - Scalability - Configuration - Module Reference - External APIs - Troubleshooting --- ## Overview Large-scale African genomic cohort studies (e.g. AWI-Gen, H3Africa, APCDR) routinely identify variants associated with complex traits such as hypertension and chronic kidney disease. Many of these variants are classified as VUS — statistically significant in the discovery cohort but lacking functional annotation to explain *how* they act biologically. This pipeline addresses the **GWAS-to-Function Gap** by integrating three independent evidence layers: 1. **AlphaGenome** (DeepMind) — deep-learning …