def parallelParse(spec, indir, run_types, run_strings):
s_mod = spec.replace(" ", "-")
spec_dir = os.path.join(args.indir, s_mod)
assert os.path.isdir(spec_dir), "\nCANNOT FIND DIRECTORY: " + spec_dir
if args.file_check:
return "", ""
# Check output directory
outline = {
"Total bases": 0,
"Total reads": 0
}
for run_ind in range(len(run_types)):
r = run_types[run_ind]
run_string = run_strings[run_ind]
run_dir = os.path.join(spec_dir, run_string)
seqfiles = mfiles.getFiles(s_mod, r, run_string, indir)
if seqfiles:
bases, reads = countReads(seqfiles, r)
outline['Total bases'] += bases
outline['Total reads'] += reads
print(outline)
if s_mod in ["Rattus-exulans", "Rattus-hoffmanni"]:
run_string += "-no-WGA"
run_dir = os.path.join(spec_dir, run_string)
seqfiles = mfiles.getFiles(s_mod, r, run_string, indir)
if seqfiles:
bases, reads = countReads(seqfiles, r)
outline['Total bases'] += bases
outline['Total reads'] += reads
return s_mod, outline
# Make the output directory for this species.
dedup_cmds = []
for r in runtype:
run_string = runstrs[r]
run_dir = os.path.join(spec_dir, run_string)
#print(run_string);
base_logfile = os.path.join(logdir,
s_mod + "-" + run_string + "-dedup")
# Get the base logfile name for this run.
seqfiles = mfiles.getFiles(s_mod,
r,
run_string,
prev_step_dir,
unmerged_flag=True)
if seqfiles:
if not os.path.isdir(run_dir):
os.system("mkdir " + run_dir)
# Make the output directory for this run.
dedup_cmds = genDedupCmd(args.path, seqfiles, r, base_logfile,
step, prev_step, dedup_cmds)
if s_mod in ["Rattus-exulans", "Rattus-hoffmanni"]:
run_string += "-no-WGA"
seqfiles = mfiles.getFiles(s_mod,
r,
run_string,
ref = "/mnt/beegfs/gt156213e/ref-genomes/mm10/mm10.fa"
elif args.ref == "-rat":
ref = "/mnt/beegfs/gt156213e/ref-genomes/rnor6/rn6.fa"
else:
ref = os.path.join(ref_dir, s_mod, s_mod + ".fa")
ref = ref.replace("..fa", ".fa")
assert os.path.isfile(ref), "\nAssembly file not found: " + ref
# Get reference
for run_ind in range(len(runtype)):
r = runtype[run_ind]
run_string = runstrs[run_ind]
#print(run_string);
run_dir = os.path.join(spec_dir, run_string)
seqfiles = mfiles.getFiles(s_mod, r, run_string, prev_step_dir)
if seqfiles:
#print(seqfiles);
#sys.exit();
if not os.path.isdir(run_dir):
os.system("mkdir " + run_dir)
bwa_cmds += genBWACmd(args.path, seqfiles, r, run_string,
s_mod, run_dir, base_logfile, ref)
if s_mod in ["Rattus-exulans", "Rattus-hoffmanni"]:
run_string += "-no-WGA"
seqfiles = mfiles.getFiles(s_mod, r, run_string, prev_step_dir)
if seqfiles:
if not os.path.isdir(run_dir):
os.system("mkdir " + run_dir)