def countOrfs(infile, outfile):
feat = "gene_id"
outfile = outfile.replace(".gz", "")
filename = re.match(
r"(\S+).(fasta$|fasta.gz|fasta.1.gz|fasta.1|fna$|fna.gz|fna.1.gz|fna.1|fa$|fa.gz|fa.1.gz|fa.1|fastq$|fastq.gz|fastq.1.gz|fastq.1)",
infile).group(1)
filemap = PipelineEnumerate.enumerateMapper(filename, PARAMS)
mapping = "sample_mappings.dir/{}/{}.mapped.bam".format(
filemap.samplename, filemap.samplename)
#get pairedness
paired = True
num_paired = subprocess.check_output(
["samtools", "view", "-c", "-f 1",
"{}".format(mapping)]).decode(sys.stdout.encoding)
if int(num_paired.strip("\n")) == 0:
paired = False
#generate counts per orf across all samples
job_threads = int(PARAMS["featureCounts_threads"])
job_memory = str(PARAMS["featureCounts_memory"]) + "G"
scratchgtf = PARAMS["featureCounts_tmp"] + filename + ".gtf"
statementlist = []
statementlist.append("zcat {} > {}".format(filemap.shortgtfpath,
scratchgtf))
statementlist.append(
PipelineEnumerate.countFeatures(feat, scratchgtf, paired, outfile,
mapping, PARAMS))
statementlist.append("gzip {}".format(outfile))
statementlist.append("rm -rf {}".format(scratchgtf))
statement = " && ".join(statementlist)
P.run(statement)
def countOrfs(infile, outfile):
feat = "gene_id"
filename = re.match(
r"(\S+).(fasta$|fasta.gz|fasta.1.gz|fasta.1|fna$|fna.gz|fna.1.gz|fna.1|fa$|fa.gz|fa.1.gz|fa.1|fastq$|fastq.gz|fastq.1.gz|fastq.1)",
infile).group(1)
filemap = PipelineEnumerate.enumerateMapper(filename, PARAMS)
mapping = "sample_mappings.dir/{}/{}.mapped.bam".format(
filemap.samplename, filemap.samplename)
#get pairedness
paired = True
num_paired = subprocess.check_output(
["samtools", "view", "-c", "-f 1",
"{}".format(mapping)]).decode(sys.stdout.encoding)
if int(num_paired.strip("\n")) == 0:
paired = False
#generate counts per orf across all samples
job_threads = int(PARAMS["featureCounts_threads"])
job_memory = str(PARAMS["featureCounts_memory"]) + "G"
statement = PipelineEnumerate.countFeatures(feat, filemap.shortgtfpath,
paired, outfile, mapping,
PARAMS)
P.run(statement)