Bowfast

Run bowtie, then BFAST on single-end colorspace reads and get a single BAM file.

Example

bowfast -f hg19.fa -p output_prefix input.csfasta input.qual

Will create output_prefix.calmd.bam which has samtools calmd -E adjusted qualities of the merged output from bowtie and BFAST

Run ./bowfast with no options to see full help.

Pipeline

automates mapping single-end colorspace reads with this pipeline:

run bowtie with conservative parameters to map high-quality reads quickly and discard reads that map to multiple locations

quality trim reads that are unmapped by bowtie

map trimmed reads with BFAST

sort and merge bowtie and BFAST outputs

index and run samtools calmd -E on merged file.

The alignments from bowtie are given fake mapping quality scores based on the number of mismatches (it's completely ad-hoc, but is tuned so that it seems to give the ability to filter).

Options

-f reference fasta file.                            [required]
-p output prefix for all files. e.g. 'data/sample1' [required]
-t number of threads.                               [default: 8]
-a option to BFAST                                  [default: 3]
   2 - keep only uniquely mapped reads (most stringent)
   3 - choose the best scoring alignment.

Installation

Note: this assumes that samtools, bowtie, and BFAST are on your path: and that indexes for bowtie and BFAST have been created. If you use the build-all.sh script in this repo, it will download and build them for you in the lib/ directory. the bowfast executable adds lib/* to your path during the execution of the script.

Download and build:

$ git clone --recursive https://github.com/brentp/bowfast/
$ cd bowfast
$ bash build-all.sh
$ ./bowfast

TODO

can add a meta script to split, parallize, and merge.
figure out trimming with sickle (see: najoshi/sickle#3)
current defaults seem to do pretty well, but can adjust -e to bowtie and -a, -b to BFAST and check results.

brentp / bowfast