PyReQTL

Index: Introduction | Features | Asking for Help | Getting Started | Operating System Requirement | Python Version | Prerequisites | Installation | Quick example | Running the Modules | Authors & Acknowledgment for the ReQTL R Toolkit | Paper | References

Introduction

PyReQTL is a python package/library that consist of collection of python modules developed for individuals or research group who prefer to perform ReQTL analysis using python instead of R programming language. Hence, this library are replica of the R toolkit "ReQTL" but implemented in Python (please see the section References for direct link of the ReQTL R toolkit GitHub repo page). The original R toolkit requires a number of R packages such as MatrixEQTL and among others Python lacks equivalent libraries. Furthermore, despite that many of the code-base in this particular project were smoothly translated into python, there were needs to interface with R and to its environment which includes importing R packages to utilize their functions from within Python and thus the Python library "rpy2" and R are requirements to these modules.

Briefly, PyReQTL does as for the ReQTL identify the correlation between expressed SNVs and their gene expression using RNA-sequencing data and each module transforms the sequencing files into ReQTL input files which will eventually be used as inputs for the R package MatrixEQTL to find the significant variation-expression relationships.

Features

• Faster than the R ReQTL version
• For efficiency and performance, it uses both Pandas DataFrames and numpy arrays
• rpy2 is more efficient and better integrated than using subprocess to interface with R

Asking for Help

If you have an issue (error or bug) when executing the library or any of the modules or have any other enquires please feel free to send me an email

Getting Started

Please follow the instructions in the sections that follow. You need to have a terminal to be able to execute the modules via the command line interface (cli). In case you are getting an error or bug please refer to the section Asking for Help.

Operating System Requirement

PyReQTL modules were developed on macOS Catalina v.10.15.2 and should work also on Linux OS.

Python Version

Both Python version 3.7 and 3.8 were successfully tested on PyReQTL. You need Python version >=3.5.

Note:

• I would recommend checking your default python version which it may be python v2.7 so running python may start version 2.6 or 2.7 and therefore you need to run python3 to start python 3 version or you may set your python v3 as your default python version.
• Note that PyReQTL were not tested on Windows and it may or may not work.

Prerequisites

• The following are the core Python libraries (dependencies) required for PyReQTL modules:

  pandas
  numpy
  scipy
  rpy2
  

• Optional Python Library:

  mypy
  

• You will also need to install the latest version of R (4.0.2) from the the official language site at: https://cloud.r-project.org

• Important notes:

  1. Python will install the required R packages for you but you may be asked to choose a secure mirror for installing those R packages.

  2. You may also be asked when executing the modules that require R packages to upgrade some R dependencies, you may ignore it by typingno.

  3. Some bioconductor packages will also be required for the bioinformatics related analysis.

Installation

1. Install the package with pip:

   pip install pyreqtl

2. You could also clone the project with the command git clone:

   git clone https://github.com/nalomran/PyReQTL.git

   • Once cloned, you may now try to install the dependencies by running the command:

       sudo pip install -r requirements.txt

   • Then, issue the ls command to list the files that end with .py (i.e. the PyReQTL modules) (OPTIONAL)

      ls -1 *.py    

Quick example

from PyReQTL import gene_matrix
gene_exp, gene_loc = gene_matrix.build_gene_matrix(gene_dir='data', prefx_out='ReQTL_test')

# file is saved for gene expression in output/ReQTL_test_gene-exp_matrix.txt
# file is saved for gene location in output/ReQTL_test_gene-exp-loc_matrix.txt

gene_exp.head(3)
# sample          gene_id  sample_1  sample_10  sample_11  sample_12  sample_13  sample_14  sample_15  ...  sample_25  sample_3  sample_4  sample_5  sample_6  sample_7  sample_8  sample_9
# 0       ENSG00000000457  0.502402  -0.502402  -0.395725  -1.020076  -0.736316  -1.768825  -1.426077  ...   1.020076  1.198380 -0.194028  1.426077 -1.198380 -0.293381  1.768825  0.096559
# 1       ENSG00000000460  0.502402   0.395725   0.000000  -1.426077  -0.502402  -0.736316  -0.869424  ...   1.020076  1.198380 -1.198380  1.426077 -1.020076 -1.768825  1.768825  0.096559
# 2       ENSG00000000938 -0.096559   1.426077  -0.293381   0.293381  -0.736316  -1.198380  -0.395725  ...   1.020076 -0.194028  1.198380  0.736316  0.194028 -1.768825  0.502402 -1.426077
# [3 rows x 26 columns]

gene_loc.head(3)
# GeneID Reference  Start     End
# 0  ENSG00000237613      chr1  34554   36081
# 1  ENSG00000238009      chr1  89295  133723
# 2  ENSG00000239945      chr1  89551   91105
# [3 rows x 4 columns])

python -m PyReQTL.gene_matrix -i data -o ReQTL_test -c True 

Running the Modules

If you choose to clone the repo from Github, then you are given two options to run the modules:

1. Running the prepared shell script "run_all.sh" found at the parent's directory of PyReQTL. This script is prepared to automatically run all modules sequentially with just a single command.

2. Running each individual python module separately.

- Option One (run all the modules sequentially)

Run the whole analysis at once:

1. type cd .. to go back to the main directory in which the shell script lives

   • You can modify the script based on your needs

2. run the following command:

sh run_all.sh

- Option Two (run each module separately)

1. build_gen_exp_matrix module

Note:

1. please refer to the link in the References section for the description of each module
2. make sure you are in the PyReQTL directory in order to run each individual python module

• Run the following command to go the PyReQTL directory:

cd PyReQTL/

Inputs + Options

• -i: a directory contain all the raw gene expression files from Stringtie

• -o: the prefix for the gene expression and gene location files

Output

• the gene expression values to be fed into MatrixEQTL

• the gene location for MatrixEQTL

Command Example

python build_gen_exp_matrix.py -i ../data -o ReQTL_test

---

2. build_VAF_matrix module

Inputs + Options

• -r: a directory containing the .csv files from the output of readCounts module

• -o: the prefix for the SNV matrix and SNV location files

Output

• the SNV locations for MatrixEQTL

• the SNV variant allele fraction matrix for MatrixEQTL

Command Example

python build_VAF_matrix.py -r ../data -o ReQTL_test

---

3. harmonize_matrices module

Input

• -v: the path to the variant matrix generated from the build_VAF_matrix.py

• -g: the path to the gene expression matrix generated from the build_gen_exp_matrix.py

• -c: the path to the covariate matrix located in the data folder

Output

• three output files consists of the three input files that have mutual samples

Command Example

python harmonize_matrices.py \
    -v output/ReQTL_test_VAF_matrix.txt \
    -g output/ReQTL_test_gene-exp_matrix.txt \
    -c ../data/covariates_matrix.txt

Note:

1. the covariates matrix file is supplied in the "data" folder
2. this is an optional module but it is recommended to avoid possible error or bug

---

4. run_matrix_ReQTL module

This module requires the R package MatrixEQTL. Please refer to the section References for more information about this package

Input

• -s: the SNV or variant matrix file created from harmonize_matrices

• -sl: the SNV location matrix file created from build_VAF_matrix

• -ge: the gene expression file matrix created from build_gene-exp_matrix

• -gl: the gene locations file created from build_gene-exp_matrix

• -c: the covariates matrix file created from "harmonize_matrices. [OPTIONAL]! you can access the file under data directory

• -o: the prefix for the path to the output files

• -ct: logical (T or F) specifying whether to split the output into cis or trans

• -pcis: p-value thresholds for the cis output files

• -ptran: p-value thresholds for the trans output files

• -p: p-value thresholds for the unified output file

Output

• one output could be cis ReQTLs and trans ReQTLs or one with all of the unified ReQTLs. This depends on what you choose for the value of parameter "-ct"

• QQ plot of p-values

Commands Example

• splitting cis and trans

python run_matrix_ReQTL.py \
      -s output/ReQTL_test_VAF_matrix_harmonized.txt \
      -sl output/ReQTL_test_VAF_loc_matrix.txt \
      -ge output/ReQTL_test_gene-exp_matrix_harmonized.txt \
      -gl output/ReQTL_test_gene-exp-loc_matrix.txt \
      -c output/covariates_matrix_harmonized.txt \
      -ct T \
      -o "ReQTL_test" \
      -pcis 0.001 \
      -ptra 0.00001

• unified cis and trans

python run_matrix_ReQTL.py \
       -s output/ReQTL_test_VAF_matrix_harmonized.txt \
       -sl output/ReQTL_test_VAF_loc_matrix.txt \
       -ge output/ReQTL_test_gene-exp_matrix_harmonized.txt \
       -gl output/ReQTL_test_gene-exp-loc_matrix.txt \
       -c output/covariates_matrix_harmonized.txt \
       -ct F \
       -o "ReQTL_test" \
       -p 0.001

---

5. annotate_cis_trans module

Input

• -r: the path to the ReQTL analysis result file

• -ga: the path to the file gene location annotations

• -o: the prefix for the output annotated result

Output

• the ReQTLs annotated as cis or trans

Commands Example

python annotate_cis_trans.py \
    -r output/ReQTL_test_all_ReQTLs.txt \
    -ga ../data/gene_locations_hg38.txt \
    -o ReQTL_test

---

Authors & Acknowledgment for the ReQTL R Toolkit

• Liam Spurr, Nawaf Alomran, Pavlos Bousounis, Dacian Reece-Stremtan, Prashant N M, Hongyu Liu, Piotr Słowiński, Muzi Li, Qianqian Zhang, Justin Sein, Gabriel Asher, Keith A. Crandall, Krasimira Tsaneva-Atanasova, and Anelia Horvath

• We would like to thank the Matrix EQTL team (Shabalin, et al. 2012) for their sample code and R package upon which run_matrix_ReQTL.R is based.

Paper

• https://academic.oup.com/bioinformatics/article/36/5/1351/5582649

References

1. The ReQTL R toolkit github repository: https://github.com/HorvathLab/ReQTL

2. MatrixEQTL R package: http://www.bios.unc.edu/research/genomic_software/Matrix_eQTL/

3. readCounts module: https://github.com/HorvathLab/NGS/tree/master/readCounts