bowtie 2 -> sorted bams

gen_bowtie2_coms.py

#!/usr/bin/env python
'''
example:
./gen_bowtie2_coms.py fastq_dir bin_dir out_dir

-all sample fastq files must be in on directory
-the bin directory should have samtools and bowtie2
-the outdir must already exist

This script prints the commands to screen.  You can save them to a file with
./gen_bowtie2_coms.py fastq_dir bin_dir out_dir > jobs.txt

'''
import sys
import re
from os import path
from collections import defaultdict as dd
from glob import glob

def abs_join(a, b):
    return path.join(path.abspath(a), b)

def get_sample_fqs(dir):
    samples = dd(lambda: {'1':[], '2':[]})
    for fi in glob(abs_join(dir, '*.fastq.gz')):
        #for each file, find its sample name and its read orientation
        mat = re.match('([\w\-_]*)_L\d\d\d_R(\d)_\d\d\d\.fastq\.gz',
                       path.basename(fi))
        s_name, pair_num = mat.groups()
        #add the file to the samples dictionary
        samples[s_name][pair_num].append(fi)
        samples[s_name]['rg_id_string'] = '--rg-id {0} --rg SM:{0} --rg LB:{0} --rg PL:ILLUMINA'.format(s_name)
    return samples

def generate_bowtie2_commands(samples, ref, bin, out, bowtie_args):
    coms = []
    for sample in samples.keys():
        com = '{0} {1} {2} -x {3} -1 {4} -2 {5} | {6} view -Su - | {6} sort - {7}; {6} index {7}.bam;'.format(
                            abs_join(bin, 'bowtie2'), 
                            samples[sample]['rg_id_string'], 
                            bowtie_args,
                            ref,
                            ','.join(samples[sample]['1']),
                            ','.join(samples[sample]['2']),
                            abs_join(bin, 'samtools'),
                            abs_join(out, sample))
        coms.append(com)
    return coms

def main():
    samples = get_sample_fqs(sys.argv[1])
    ref = sys.argv[2]
    bindir = sys.argv[3]
    outdir = sys.argv[4]
    try:
        bowtie2_args = sys.argv[5]
    except IndexError:
        bowtie2_args = '--no-unal --very-sensitive-local'

    commands = generate_bowtie2_commands(samples, ref, bindir, outdir, bowtie2_args)
    print '\n'.join(commands)

if __name__ == '__main__':
    main()