PreComputeCapice

When you want the full capice experience, but don't want to use the model

This is a program that works on the wonderfull CAPICE predictive model.

Introduction

'PreComputeCapice' is a program developed by R.J. Sietsma and maintained by the Genomic Coordination Centre Groningen. The program calculates capice scores for all entries in a given CADD file.

Prerequisites

This program is developed in Pycharm 2020.1 (Professional Edition), performance on other systems is not guaranteed.

The program requires the following packages:

• numpy (v1.18.3; BSD 3-Clause License)
• pandas (v1.0.3; BSD 3-Clause License)
• psutil (v5.7.0; BSD 3-Clause License)
• scipy (v1.4.1; BSD 3-Clause License)
• scikit-learn (v0.19.1; BSD 3-Clause License)
• xgboost (v0.72.1; Apache 2 License)

Warning: this program works for python version 3.6, it does not work for python 3.7 or higher because of numpy, scikit-learn and xgboost version dependency issues.

Installing

Step 1: acquire the source files Either clone or download the source files.

Step 2: activate the virtual environment

• Open a terminal in the cloned or downloaded folder.
• Make sure you have python3.6 installed by typing python3 --version.
• Execute the following:

mkdir ./venv
cd ./venv
python3 -m venv ./
cd ..

• Activate the virtual environment:

source ./venv/bin/activate

• Install the required packages by executing:

pip install -r requirements.txt

Note: if any package fails to install, please try to install the package using:

pip install package==version

Make sure you have the virtual environment enabled before you install packages!

Usage

The program requires the following arguments:

• -f / --file: the cadd file in gzip (.gz) format.
• -m / --model: the pickled capice model in .dat format.
• -o / --output: the location where the program should place it's files.

Optional argument:

• -s / --batchsize: the amount of rows the program should read each iteration from the CADD file.

Example usage:

with a batch size of 1 million

python3 PreComputeCapice.py -f path/to/cadd/file.gz -m path/to/model.dat -o path/to/output/folder -s 1000000

Output

The program will output the following files:

• For each chromosome in the CADD file, it makes a folder named chrx (where x = chromosome) and places a gzipped tsv of all CADD entries for that chromosome. Note: The program continually adds entries to this file, do NOT remove or replace this file till the program is done!
• Log_output: a file with timed messages on updates within the program. (Does not contain error messages or warnings).
• progression_json: a json file containing set parameters, like batch_size, to keep track of progress during the programs execution.

TODO:

• Make input file (-f / --file) also specific for the progression.json.
• Refactoring and optimization.