PBVI

PBVI Position-Based Variant Identification (PBVI) is a method used to identify low-level variants by directly modeling sequencing error from a control dataset.

This repository contains model files and scripts that can be used to detect low-level variants in 11 overgrowth syndrome genes. It also contains custom scripts used to assess the sensitivity of this approach.

The PBVI scripts are written in Python 2 (version 2.7) and are not compatible with Python 3.

Overview of PBVI

The PBVI method involves two steps. First, a control population dataset is used to generate a model that can be used for variant calling. Second, variants in input BAM files are called using Fisher’s exact test.

1. Model building

The model can be built by modifying variables (see below) in the generate_OGPB_model.py script. The following data are required for this step:

• Control population BAM files: the BAM files with which to build the model.
• BED file: the genomic locations to examine.

> bamfiles = glob.glob('/BAM_FILE_DIRECTORY/*.bam') #CHANGE THIS TO THE DIRECTORY WHERE BAM FILES ARE, this will look for *bam files
> bedfile = 'OGPB_model_location.bed' #CHANGE THIS TO INPUT BEDFILE

The generate_OGPB_model.py script will create a data frame of aggregated counts and output this to a MODEL.txt file.

2. Variant calling

After building the model, the PBVI.py script can be modified (see below) to call variants in input BAM files. The following data are required for this step:

• Input BAM files: the BAM files in which to call variants.
• BED file: the BED file used to build the model.
• Reference genome file: a FASTA reference file.
• Model text file: the MODEL.txt file output from the generate_OGPB_model.py script.

> bamfiles = glob.glob('/BAM_FILE_DIRECTORY/*.bam') #CHANGE THIS TO THE DIRECTORY WHERE BAM FILES ARE, this will look for *bam files
> bedfile = 'OGPB_model_location.bed' #CHANGE THIS TO INPUT BEDFILE
> fastafile = 'hg19.fa' #CHANGE THIS TO FASTA REFERENCE FILE
> posmodelfile = 'OGPB_model.txt' #CHANGE THIS TO MODEL FILE

For each input BAM file, the PBVI.py script will output a text file containing the called variants.

The generate_OGPB_model.py and PBVI.py scripts are available in PBVO-OGPB directory.