def merge(assembled_gtfs, ref_file, gtf_file, num_cores, data):
"""
run cuffmerge on a set of assembled GTF files
"""
assembled_file = tempfile.NamedTemporaryFile(delete=False).name
with open(assembled_file, "w") as temp_handle:
for assembled in assembled_gtfs:
temp_handle.write(assembled + "\n")
out_dir = os.path.join("assembly", "cuffmerge")
merged_file = os.path.join(out_dir, "merged.gtf")
out_file = os.path.join(out_dir, "assembled.gtf")
if file_exists(out_file):
return out_file
if not file_exists(merged_file):
with file_transaction(data, out_dir) as tmp_out_dir:
cmd = ("cuffmerge -o {tmp_out_dir} --ref-gtf {gtf_file} "
"--num-threads {num_cores} --ref-sequence {ref_file} "
"{assembled_file}")
cmd = cmd.format(**locals())
message = ("Merging the following transcript assemblies with "
"Cuffmerge: %s" % ", ".join(assembled_gtfs))
do.run(cmd, message)
clean, _ = clean_assembly(merged_file)
fixed = fix_cufflinks_attributes(gtf_file, clean, data)
classified = annotate_gtf.annotate_novel_coding(fixed, gtf_file, ref_file,
data)
filtered = annotate_gtf.cleanup_transcripts(classified, gtf_file, ref_file)
shutil.move(filtered, out_file)
return out_file
def test_annotation():
safe_makedir("test_gtf")
fixed = annotate_gtf.annotate_novel_coding(TEST_ASSEMBLED, TEST_GTF,
TEST_FASTA)
cleaned = annotate_gtf.remove_noncongruent_transcripts(
fixed, TEST_GTF, TEST_FASTA)
print cleaned
def merge(assembled_gtfs, ref_file, gtf_file, num_cores, data):
"""
run cuffmerge on a set of assembled GTF files
"""
assembled_file = tempfile.NamedTemporaryFile(delete=False).name
with open(assembled_file, "w") as temp_handle:
for assembled in assembled_gtfs:
temp_handle.write(assembled + "\n")
out_dir = os.path.join("assembly", "cuffmerge")
merged_file = os.path.join(out_dir, "merged.gtf")
out_file = os.path.join(out_dir, "assembled.gtf")
if file_exists(out_file):
return out_file
with file_transaction(data, out_dir) as tmp_out_dir:
cmd = ("cuffmerge -o {tmp_out_dir} --ref-gtf {gtf_file} "
"--num-threads {num_cores} --ref-sequence {ref_file} "
"{assembled_file}")
cmd = cmd.format(**locals())
do.run(cmd, "Merging transcript assemblies with reference.")
clean, _ = clean_assembly(merged_file)
fixed = fix_cufflinks_attributes(gtf_file, clean, data)
classified = annotate_gtf.annotate_novel_coding(fixed, gtf_file, ref_file)
filtered = annotate_gtf.cleanup_transcripts(classified, gtf_file, ref_file)
os.rename(filtered, out_file)
return out_file
def test_annotation():
safe_makedir("test_gtf")
fixed = annotate_gtf.annotate_novel_coding(TEST_ASSEMBLED, TEST_GTF, TEST_FASTA)
cleaned = annotate_gtf.remove_noncongruent_transcripts(fixed, TEST_GTF, TEST_FASTA)
print cleaned
