Genomic Analysis Pipeline - GEMMA

Overview

This repository contains bash scripts for genomic variant association analysis. All specific file paths have been replaced with configurable environment variables for portability and security.

Setup Instructions

1. Configure Paths

Edit config_paths.sh to set your environment-specific paths:

export BASE_DIR="/path/to/base/directory"
export IMPUTATION_DIR="/path/to/imputation/directory"
export PROJECT_DIR="/path/to/project/directory"

2. Source Configuration

Before running any analysis script, source the configuration file:

source config_paths.sh

3. Create Directory Structure

Create all necessary directories:

source config_paths.sh
create_directories

Path Mapping Reference

Variable	Description	Original Path
BASE_DIR	Base directory for original genotype data	NA
IMPUTATION_DIR	Imputated data directory	NA
PROJECT_DIR	Project root directory	NA
DATA_DIR	Data directory (relative)	NA
PHENO_FILE	Phenotype file	NA
SEX_FILE	Sex update file	NA

Analysis Pipeline

Sample Cleanup (cmd05)

source config_paths.sh
bash cmd05_sampleCleanup.sh

• Filters samples and removes duplicates
• Identifies samples with high genotype missing rate

MDS Analysis (cmd06)

source config_paths.sh
bash cmd06_mds.sh
bash cmd06_2_mds_ocm50k.sh

• Calculates MDS eigenvalues for pooled and population-specific data
• Generates IBD estimates
• Performs population stratification analysis

GEMMA GWAS - Array Data (cmd07)

source config_paths.sh
bash cmd07_gemma_array_1_GRM.sh      # Generate GRM
bash cmd07_gemma_array_2_maf_run.sh  # Run GEMMA
bash cmd07_gemma_array_3_plot.sh     # Plot results

• Filters to 711 samples with valid phenotypes
• Generates genetic relationship matrices
• Runs GWAS with MAF thresholds (0.01, 0.05)
• Adjusts for sex and principal components
• Generates Manhattan and QQ plots

GEMMA Association - OCM50k Region (cmd09)

source config_paths.sh
bash cmd09_gemma_impuOCM50k_1_dataPrep.sh  # Prepare data
bash cmd09_gemma_impuOCM50k_2_run.sh       # Run GEMMA

• Focuses on OCM 50kb region

R Scripts

one line summary

• r_scp_Gemma_dataPrep_cov_array.R: Builds GEMMA covariate files (intercept + sex + selected MDS PCs) for the array genotype dataset.
• r_scp_Gemma_dataPrep_cov_oriOCM50k.R: Builds GEMMA covariate files (intercept + sex + selected MDS PCs) for the original OCM50k dataset.
• r_scp_plot_GEMMA_gwas_arg.R: Generates QQ and Manhattan plots from a GEMMA GWAS association output specified via command-line arguments.
• r_proc_plot_GEMMA_OCM50k.R: Produces QQ and Manhattan plots for GEMMA results in the OCM50k region, including an alternative labeled Manhattan plot.

Configuration (paths) for R script

All cleaned scripts remove hard-coded absolute paths and instead use environment variables:

• SAMOCM_PROJECT_DIR: directory where these scripts live (default: current working directory).
• SAMOCM_Analyst_DATA_DIR: path to Analyst``_data (default: file.path(SAMOCM_PROJECT_DIR, '..', '``Analyst``_data')).

Key Output Files

Genetic Relationship Matrices (GRM)

• ${GEMMA_ARRAY_DIR}/*standrel.sXX.txt

GWAS Results

• ${GEMMA_ARRAY_DIR}/gemma_out_*.assoc.txt - Array-based GWAS
• ${GEMMA_IMPU_DIR}/gemma_out_*.assoc.txt - Imputation-based GWAS
• ${GEMMA_OCM50K_DIR}/gemma_out_*.assoc.txt - OCM50k region GWAS

MDS/PCA Results

• ${MDS_DIR}/*_mds.mds - MDS coordinates

Sample Lists

• ${LIST_DIR}/list.id.rm - Samples to remove (n=19)
• ${LIST_DIR}/list.id.JAM - Jamaican samples
• ${LIST_DIR}/list.id.MAL - Malawian samples

Module Requirements

module load plink/1.9.0-beta4.4
module load GEMMA/0.96
module load R

Analysis Parameters

MAF Thresholds

• 0.01 (1% minor allele frequency)
• 0.05 (5% minor allele frequency)

Covariate Models

• SEX - Sex only
• SEX_C1 - Sex + PC1
• SEX_C1-C2 - Sex + PC1-2
• SEX_C1-C3 - Sex + PC1-3

LD Pruning

• Window: 50 SNPs
• Step: 10 SNPs
• r² threshold: 0.1

Sample Sizes

• Final analysis: 711 samples
  • Jamaican (JAM): 340 samples
  • Malawian (MAL): 371 samples

Citation

If using these scripts, please cite the original study and acknowledge the data sources.

Contact

For questions about the analysis pipeline, please contact the study investigators.