def run(self, outfile, params):
path = os.environ["PATH"]
gp = P.get_parameters_as_namedtuple()
cluster_queue = gp.cluster["queue"]
cluster_memory_resource = gp.cluster["memory_resource"]
cluster_parallel_environment = gp.cluster["parallel_environment"]
outdir = os.path.dirname(outfile)
outname = os.path.basename(outdir)
# -sync y forces qsub to wait until job completes before
# continuing.
statement = (
"{self.path} "
"-p canu "
"-d {outdir} "
"-genomeSize={params.genome_size} "
"gridOptionsJobName={outname} "
"java={params.path_java} "
"gridOptions=\"-q {cluster_queue} -v PATH={path} -sync y \" "
"gridEngineMemoryOption=\"-l {cluster_memory_resource}=MEMORY\" "
"gridEngineThreadsOption=\"-pe {cluster_parallel_environment} THREADS\" "
"{params.options} "
"{params.assembly_mode} "
"{params.fasta} "
">& {outfile}.log; "
"mv {outdir}/canu.contigs.fasta {outfile}".format(**locals()))
return P.run(statement, without_cluster=True)
def pre_process(self, infile, outfile, params):
statements = []
infile = IOTools.snip(infile, ".bam")
tmpdir = P.get_parameters_as_namedtuple().tmpdir
outprefix = os.path.basename(os.path.dirname(outfile))
if params.copy_bam:
statements.append("cp @[email protected] @[email protected]; "
"cp @[email protected] @[email protected]")
if params.split_bam:
statements.append("daisy bam2bam-split-reads "
"-i @[email protected] "
"-o - "
"{params.split_bam} "
"--log={outfile}_split_bam.log "
"2> {outfile}_split_bam.err "
"> @[email protected]; ".format(**locals()))
if params.bam2bam:
statements.append("daisy bam2bam "
"--stdin=@[email protected] "
"{params.bam2bam} "
"--log={outfile}_bam2bam.log "
"2> {outfile}_bam2bam.err "
"> @[email protected]; ".format(**locals()))
if params.region:
statements.append(
"samtools view -b @[email protected] {} > @[email protected]".format(
params.region))
if params.shift_quality:
statements.append("samtools view -h @[email protected] "
"| perl -lane "
"'if(/^@/) {{print; next;}} "
"@qual=split(//, $F[10]); "
"$_=chr(ord($_)+{}) for (@qual); "
"$F[10]=join(\"\",@qual); "
"print join(\"\\t\", @F)' "
"| samtools view -bS > @[email protected]".format(
params.shift_quality))
if is_true(params.remove_chr):
# also substitute chrM to MT.
statements.append("samtools view -h @[email protected] "
"| awk -v OFS='\\t' '"
"$1 == \"@SQ\" "
"{{ gsub(\"chrM\", \"chrMT\", $2); "
" gsub(\"chr\", \"\", $2); print; next }} "
"{{ gsub(\"chrM\", \"chrMT\", $3); "
" gsub(\"chr\", \"\", $3); print; next}} '"
"| samtools view -bS - "
"2> {outfile}_remove_chr.log "
"> @[email protected]; ".format(**locals()))
if not statements:
return infile + ".bam", "", ""
filename, build_statement, cleanup_statement = P.join_statements(
statements, infile)
filename += ".bam"
build_statement += (
"; samtools index {filename} >& {outfile}.index.log".format(
**locals()))
return filename, build_statement, cleanup_statement