def filterAndTrim(infile, outfile):
'''
filter and trim reads based on user-defined
parameters
'''
maxn = PARAMS["trim_maxn"]
maxee = PARAMS["trim_maxee"]
truncQ = PARAMS["trim_truncq"]
truncLen = PARAMS["trim_trunclen"]
trimLeft = PARAMS["trim_trimleft"]
# make sure parameters are present
assert maxn != "" and maxee != "" and truncQ != "" and truncLen != "" and trimLeft !="", \
"must specify all parameters to filterAndTrim"
if PARAMS["paired"] == 1:
paired = "--paired"
# unzip one by one
infile_read2 = infile.replace(".fastq.1.gz", ".fastq.2.gz")
infiles = [infile, infile_read2]
else:
paired = ""
infiles = [infile]
tmpdir = P.get_temp_dir()
for inf in infiles:
outtmp = os.path.join(tmpdir, inf.replace(".gz", ""))
statement = '''zcat %(inf)s > %(outtmp)s'''
P.run(statement)
# hackabout
outtmp = os.path.join(tmpdir,
[x for x in infiles
if x.endswith(".fastq.1.gz")][0].replace(".gz", ""))
statement = '''Rscript %(scriptsdir)s/dada2_filter_and_trim.R
--infile=%(outtmp)s
%(paired)s
--maxN=%(maxn)s
--maxEE=%(maxee)s
--truncQ=%(truncQ)s
--truncLen=%(truncLen)s
--trimLeft=%(trimLeft)s
--filtered-directory=filtered.dir'''
P.run(statement)
shutil.rmtree(tmpdir)
