reaganmogire/lft-gwas-afr
Domaine:
healthcareCréateur:
rea
Hôte:
Template scripts for multi-cohort GWAS of liver enzyme traits in African populations
# Multi-cohort GWAS of Liver Enzyme Traits (ALP, ALT, AST, GGT)
This repository contains **minimal, cohort-agnostic template scripts** illustrating the main analytical steps used for genome-wide association and downstream statistical genetics analyses of circulating liver enzyme traits. The goal is to document **methods and execution patterns**; it is not a drop-in, end-to-end reproduction of any single cohort pipeline.
The repository accompanies the manuscript:
*Multi-cohort genome-wide association analyses reveal loci underlying circulating liver enzyme levels in African-ancestry populations.*
No individual-level data or cohort-specific configuration files are included.
---
## Repository scope
The repository documents the following analytical components:
1. Genome-wide association analyses (**SAIGE**)
2. Meta-analysis across cohorts (**METASOFT**, Han–Eskin random effects)
3. Conditional analyses (**GCTA–COJO**)
4. Statistical fine-mapping (**SuSiE RSS**)
5. Colocalisation with cis-eQTLs (**FastENLOC**, GTEx liver)
6. Secondary colocalisation using a signal-isolated liver eQTL resource (**COLOC / coloc.abf**)
7. Genetic correlation analyses (**LD score regression**)
Each component is represented by a single template script and/or an example command.
---
## End-to-end analysis outline
**Step 1 — Run within-cohort GWAS (SAIGE).**
Fit a null model per phenotype and cohort, then run association testing per chromosome.
**Step 2 — Harmonize cohort GWAS summary statistics.**
Ensure consistent variant identifiers, alleles, effect direction, and column schema across cohorts.
**Step 3 — Meta-analyze across cohorts (METASOFT).**
Perform Han–Eskin meta-analysis per phenotype using harmonized cohort-level summary statistics.
**Step 4 — Test independence of “novel” signals (GCTA–COJO).**
Condition lead signals on previously reported variants within the locus using an ancestry-matched LD reference.
**Step 5 — Fine-map selected loci (SuSiE RSS).**
Construct locus- …