Multi-ancestry analysis of plasma protein levels influencing and responding to major depression liability

Over a year ago through an MRC Network Grant researchers at the University of Bristol visited Entebbe, Uganda to provide a course on genetic epidemiology / Mendelian randomisation. Part of that course involved a hackathon to put those skills to use and this work is an extension of Hackathon group 6 - MR course, Uganda, 2023.

Objectives

Estimate the influence of pharmacological targets for incidence of major depressive disorder (MDD) across multiple ancestries using Mendelian randomisation

Forward MR Analysis

1. Download data - MDD GWASs

./scripts/download_mdd.sh

2. Identify proteins relevant to MDD in Europeans. Slight challenge - pQTL data is chr:pos, but MDD data is rsid. Need to map rsid to chr:pos using dbSNP, and then lookup pQTLs in MDD GWAS.

Need to have dbSNP VCF downloaded, and bcftools on the PATH.

2.1. Map MDD data to chr:position
2.2. Lookup cis-pQTLs in MDD data
2.3. Identify pQTL effects in MDD with FDR < 0.05
2.4. Write out list of proteins that have a cis effect in MDD

Rscript scripts/identify_pqtls.r

This identifies 17 proteins to use in the multi-ancestry comparison for effects on MDD.

3. The Eur pQTLs are ascertained for high LD tagging in Europeans only, so it will disadvantage power in non-Europeans. Use the cross-ancestry pQTL meta analysis to analyse the region around a pQTL. Choose the best SNP in the region for each of the 17 pQTLs.

Rscript scripts/cross_ancestry_pqtl_selection.r

4. Extract the ancestry-agnostic pQTL SNPs from each of the MDD GWASs

Rscript scripts/extract_mdd.r

5. Perform per ancestry MR analysis EAS: scripts/EAS_analysis_modified_Forward_MR.r AFR: scripts/afr_analysis_modified_Forward_MR.R EUR: scripts/eur_analysis_modified_Forward_MR.r SAS: scripts/csa_sas_modified_Foward_MR.r

6. Perform combined MR and Heterogeneity analysis Use scripts/combined_analysis.qmd

Reverse MR

1. Generated bim file with scripts/bim_for_prs script.
2. Calculate per ancestry PRS using PRScsx.py (refer to PRScsx), MDD per ancestry GWAS summary stats and bim file using scripts/PRScsx.sh .
3. Use scripts/PRS/Association_Analysis.Rmd to perform MDD PRS-protein association and heterogenity analysis.

Foward MR for known druggable targets

1. Use script scripts/identify_druggable_eqtls_Eur_ancestry to obtain IVs for EUR and scripts/identify_druggable_eqtls_All_ancestry script for all ancestry GWAS data. AFR, SAS and EAS have no significant Ivs

2. Perform per ancestry MR using rmd files EAS:scripts/Foward_mr_druggable_eas.Rmd AFR: scripts/Foward_mr_druggable_afr.Rmd EUR: scripts/Foward_mr_druggable_eur.Rmd SAS: scripts/Foward_mr_druggable_sas.Rmd

3. Perform combined MR and heterogenity analysis using scripts/combined_analysis _druggable_MR

MDD liability prediction

Use scripts\protein-MDD-liability-prediction.Rmd to assess the ability of reverse MR in identifying predictive biomarkers of MDD onset.