CombiVEP 0.1.2

Combined Variant Effect Predictors

Pre-requisite

• Python 2.7

• pip (http://pypi.python.org/pypi/pip)

• numpy (http://pypi.python.org/pypi/numpy)

• matplotlib (http://matplotlib.org)

Installation

To install this module, run the following commands:

sudo python setup.py install

Download reference database

After the installation is complete, type

CombiVEP_reference_updater

This application will automatically check with UCSC and LJB database, and see if it is required to download the new one. The original database size is around 1GB each. The total operation time for each database should be around 30-60 mins.

Training

After having reference database installed, the CombiVEP model can be trained using

CombiVEP_trainer <training_data_file>

<training_data_file> must be in CBV format: CHROM, POS, REF, ALT, ACTUAL_DELETERIOUS_EFFECT. Each field is separated by a tab. SNP Position(POS) is 1-based index. The VariBench training data file in CBV format can be found at

combivep/data/CBV/training.cbv

Prediction

To use the trained model to predict the effect, you can do it using

CombiVEP_predictor <input_file> [-F FORMAT]

The input file can be either in VCF or above CBV format. Default is in VCF format. So if you want to use input file in VCF format, simply type

CombiVEP_predictor <vcf_file>

If you to do the prediction using file in CBV format, you can

CombiVEP_predictor <cbv_file> -F CBV

The VariBench test data file in CBV format can be found at

combivep/data/CBV/test.cbv