PMpred 1.0.4

A Python package that adjusts GWAS summary statistics for the effects of Sparse Precision Matrix (PM)

PMpred

PMpred is a Python based software package that adjusts GWAS summary statistics for the effects of precision matrix (PM), which is the inverse of linkage disequilibrium (LD).

• The current version is 1.0.0

Getting Started

PMpred can be installed using pip on most systems by typing

pip install pmpred

Requirements

LDpred currently requires three Python packages to be installed and in path. These are numpy https://numpy.org/, scipy http://www.scipy.org/ and joblib https://joblib.readthedocs.io/en/stable/. Lastly, PMpred has currently only been tested with Python 3.6+.

The first two packages numpy and scipy are commonly used Python packages, and pre-installed on many computer systems. The last joblib package can be installed using pip https://joblib.readthedocs.io/en/stable/, which is also pre-installed on many systems.

With these three packages in place, you should be all set to install and use PMpred.

Installing PMpred

As with most Python packages, configurating LDpred is simple. You can use pip to install it by typing

pip install pmpred

This should automatically take care of dependencies. The examples below assume ldpred has been installed using pip.

Alternatively you can use git (which is installed on most systems) and clone this repository using the following git command:

git clone https://github.com/WiuYuan/pmpred.git

Then open the terminal of the repository folder and run command:

pip install .

Finally, you can also download the source files and place them somewhere.

With the Python source code in place and the three packages numpy, scipy, and joblib installed, then you should be ready to use PMpred.

Using PMpred

A typical LDpred workflow consists of 3 steps:

Step 1: Get data incude Precision Matrix, Snplists and GWAS Sumstats

The first step is to prepare the data we use in PMpred, contain {Precision Matrix, Snplists, GWAS Sumstats}

• Precision Matrix: could be download in https://zenodo.org/records/8157131
• Snplists: could be download in https://zenodo.org/records/8157131
• GWAS Sumstats: should be prepared using csv format with split \t and need include five head {rsid, REF, beta, beta_sd, N}. An example is showed below:

rsid    REF    ALT    beta    beta_sd    N ...
  *      *      *        *       *       *
  *      *      *        *       *       *
  *      *      *        *       *       *
...

Certainly, you can specify the headers of sumstats and split with parameters in pmpred like below:

--rsidname SNP
--REFname A1
--ALTname A2
--betaname BETA
--sename SE
--Nname n
--split ,
...

Then the sumstats could be like:

SNP,A1,A2,BEAT,SE,N,...
*,*,*,*,*,*,...
*,*,*,*,*,*,...
...

Step 2: Choose the method using in PMpred

After getting the required data we could easily get the effect size using the quick start below:

pmpred --pm precision_matrix_folder --snp snplists_folder -s sumstats_file -o output_file

If you use precision matrix many times, you could first normalize it using command below:

pmpred --pm precision_matrix_folder -o new_precision_matrix_folder

then use pmpred without normalize Precision Matrix

pmpred --pm precision_matrix_folder --snp snplists_folder -s sumstats_file -o output_file --unnormal

Other parameters in PMpred could be found in

pmpred -h