rowbowt: run-length BWT tools for working with genomic sequences

Author: Taher Mun

Code heavily derives from Nicola Prezza's r-index repository.

Compiling

Requires a c++17 compliant compiler and a Unix-based operating system.

To download and build:

git clone github.com/alshai/rowbowt;
mkdir rowbowt/build;
cd rowbowt/build;
cmake ..
make

This produces the exectuables rb_align and rb_build.

Building a rowbowt index

1. generate files needed for building the index (exact format specifications will be added below):

   • .bwt (required) - contains the BWT of the original text. This will let you find the number of occurences of your pattern in the original text

   • .ssa, .esa (optional) - contains the run-starts and run-ends SA samples, respecitively. Use this if you want use the index to find pattern locations in the original text.

   • and .ma files - contains the markers (ref position + allele) associated with each SA element. Use this if you want to find what alleles wrt the reference sequence your pattern overlaps.

   • we recommend using pfbwt-f (included in this repo) to generate these required files from a reference sequence and a VCF containing a haplotype reference panel. See the README for pfbwt-f for details on how to run.

   cd pfbwt-f
   make vcf_to_bwt
   python3 vcf_to_bwt.py [-s] [-m] -o <output prefix> <reference fasta> <VCF>
   

2. Run the build command

   ./rb_build [-s] [-m] -o <output prefix> <rowbowt index prefix> 
   

Querying sequences against a rowbowt index

./rb_align [-s] [-m] <rowbowt index prefix> <fastq>

• use -s and -m to query locations and markers, respectively.

• Proper formatting (ie. sam format for locations, VCF format for markers) will be added in the future.