Single cell co-expression QTL analysis

This repository contains the code to generate the results and figures from

Identification of genetic variants that impact gene co-expression relationships using large-scale single-cell data

Shuang Li *, Katharina T. Schmid *, Dylan de Vries **, Maryna Korshevniuk **, Corinna Losert, Roy Oelen, Irene van Blokland, BIOS Consortium, sc-eQTLgen Consortium, Hilde E. Groot, Morris A. Swertz, Pim van der Harst, Harm-Jan Westra, Monique van der Wijst, Matthias Heinig †, Lude Franke †

* These authors contributed equally
** These authors contributed equally
† These authors contributed equally

Publication (Genome Biology, 2023): https://doi.org/10.1186/s13059-023-02897-x

Preprint: https://www.biorxiv.org/content/10.1101/2022.04.20.488925v1

Overview

The code for the analysis is separated in different steps, each in its own subdirectory including a README file:

• Exploring different different association metrics and other GRN construction tools, including pseudotemporal based ones and combing cells to meta cells 01_association_metrics/
• Comparing correlation between the different single cell data sets, with bulk data and CRISPR knock-out data; testing potential occurence of Simpson's paradox 02_correlation_evaluation/
• Comparing correlation between different cell types and between different individuals within one cell type 03_celltype_individual_comparison/
• Running eQTL and coeQTL mapping pipelines followed by replication in bulk and technical evaluation of the co-eQTLs, such as correlation distribution, sub cell type effects and effects of subsampling cells or donors 04_coeqtl_mapping/
• Interpretation of co-eQTL results based on GWAS annotation and different enrichment analyses (GO enrichment, TFBS enrichment using Remap database and GWAS enrichment using MAGMA) 05_coeqtl_interpretation/

Software requirements

Most code is implemented in R or python, the required packages are documented in the respective yaml files (conda_env_R.yml and conda_env_python.yml) and can be used to setup the respective conda environments:

conda env create -f conda_env_R.yml
conda activate r_env

The following R packages are not part of conda and need to be added afterwards (if the respective code part should be run):

if (!require("BiocManager")) install.packages('BiocManager') 
BiocManager::install("tanaylab/metacell")

devtools::install_github("heiniglab/scPower")

Further external tools were used:

• To calculate the eQTLs See the documentation here.
• To calculate the co-eQTLs: mbQTL.
• To perform GWAS enrichment analysis on the co-eQTL genes: MAGMA.