subsample fastq reads

enrichment_sub.py

#!/usr/bin/env python

#subsampling and read mapping
#for script information/arguments: "python enrichment_sub.py --help"

import sys
import glob
from subprocess import Popen, PIPE
import random
import argparse
import os

#To parse command line
parser = argparse.ArgumentParser()

parser.add_argument("-d", "--directory", type=str, help="directory of fastqs")
parser.add_argument("-i", "--iterations", type=int, help="number of interations")
parser.add_argument("-s", "--samplingdepth", type=int, help="number of reads to subsample",)

args = parser.parse_args()

fastqs = glob.glob("%s*_R1_*fastq.gz" % (args.directory))

def is_tool(name):
    from distutils.spawn import find_executable
    return find_executable(name) is not None

result = is_tool("seqtk")

if result is False:
    print("seqtk is not installed but needs to be...exiting")
    sys.exit()

for file in fastqs:
    seeds = random.sample(range(1000), args.iterations)
    for seed in seeds:
        R2file="_R2_".join(file.split("_R1_"))
        outfastqR1= "%d_%s" % (seed, file.strip(args.directory))
        outfastqR2= "%d_%s" % (seed, R2file.strip(args.directory))
        os.system("seqtk sample -s%d %s %d | gzip > %s" % (seed, file, args.samplingdepth, outfastqR1))
        os.system("seqtk sample -s%d %s %d | gzip > %s" % (seed, R2file, args.samplingdepth, outfastqR2))