mrpeg

mrpeg is a Python software to perform association test integrating perturbational screens, eQTL, and GWAS summary data to identify mediating genes of complex traits.

- We detest usage of our software or scientific outcome to promote racial discrimination.

The Mr. PEG manuscript is described in

Integrating perturbational screens, eQTL, and GWAS data identifies mediating genes for complex traits

Zeyun Lu, Yi Ding, Nathan LaPierre, Lili Wang, Douglas Yao, Nicholas Mancuso, Alexander Gusev

Check here for full documentation.

Installation | Example | Version History | Support

Installation

1. Before installation, we highly recommend to create a new environment using conda so that it will not affect the software versions of the other projects. For example, use following codes:

   conda create -n env-mrpeg python=3.8

   We currently only support Python3.8+.

2. If you are using a Mac with an Apple M1 or newer chip, you should install cbgen package or other required pacakges from conda-forge first to ensure compatibility (see this link for previous issue). One easy workaround is to initiate your conda using miniforge. On most HPC systems, this is usually not necessary.

   conda install -c conda-forge cbgen

3. Last, users can download the latest repository and then use pip:

   git clone https://github.com/gusevlab/mrpeg.git
   cd mrpeg
   pip install .

Get Started with Example

mrpwg software is very easy to use:

For performing inference:

cd ./data/
mrpeg peg --gwas example_gwas.tsv.gz \
  --eqtl example_eqtl.tsv.gz \
  --perturb example_perturbation.tsv.gz \
  --gwas-cols chrom snp a1 a0 beta se \
  --eqtl-cols chrom snp a1 a0 z gene \
  --ref-geno plink/geno_chr\* \
  --trait "mediating_gene" \
  --top-signal 1 \
  -o tmp_results_mediating_gene

We also implement two functions for gene annotation.

1. Find the GWAS closest genes:

cd ./data/
mrpeg closest --gwas example_gwas.tsv.gz \
  --gwas_cols chrom snp pos beta se \
  --ref ref_gene_info.tsv.gz \
  --ref_cols CHR TSS TES ID2 \
  --trait example \
  -o tmp_results_closest

2. Compute the GWAS signals given gene annotations.

cd ./data/
mrpeg signal --gwas example_gwas.tsv.gz \
  --gwas_cols chrom snp pos beta se \
  --ref ref_gene_info.tsv.gz \
  --ref_cols CHR P_MID_FLANK0 P_MID_FLANK1 ID2 \
  --window 0 \
  --trait example \
  --chr 1 \
  -o tmp_results_signal

See here for more details on how to use mrpeg.

If you want to use in-software mrpeg inference function, you can use following Python code as an example:

from mrpeg.peg import infer_peg
# betas is a numpy array of GWAS effects (k by 1)
# ses is a numpy array of GWAS standard errors (k by 1)
# eqtl is a numpy array of eQTL effects (k by 1)
# perturbs is a numpy array of perturbation effects (k by t) of k perturbed genes on t downstream genes
# ld is a numpy array (k by k) of LD across SNPs
infer_peg(beta=betas, se=ses, eqtl=eqtls, perturb=perturbs, ld=lds)

You can customize this function with your own ideas!

Version History

Version	Description
0.1	Initial Release

Support

For any questions, comments, bug reporting, and feature requests, please contact Zeyun Lu ([email protected]) and Sasha Gusev ([email protected]), and open a new thread in the Issue Tracker.

---

This project has been set up using PyScaffold 4.1.1. For details and usage information on PyScaffold see https://pyscaffold.org/.