def test_standarize_sam():
'It test that we can add default qualities to the sanger reads'
sam_fhand = StringIO('''@SQ\tSN:SGN-U576692\tLN:1714
@SQ\tSN:SGN-U572743\tLN:833
@RG\tID:g1\tLB:g1\tSM:g1\tPL:sanger
@RG\tID:g3\tLB:g3\tSM:g3\tPL:sanger
SGN-E200000\t64\tSGN-U572743\t317\t226\t254M24S\t*\t0\t0\tGGATGATKTTAGAG\t*\tAS:i:250\tXS:i:0\tXF:i:0\tXE:i:7\tXN:i:0\tRG:Z:g1
SGN-E40000\t0\tSGN-U576692\t1416\t207\t168M\t*\t0\t0\tAGCCTGATAA\t,,09377777\tAS:i:160\tXS:i:0\tXF:i:3\tXE:i:4\tXN:i:0\tRG:Z:g3
SGN-E40000\t20\tSGN-U576692\t1416\t207\t168M\t*\t0\t0\tAGCCTGATAA\t,,09377777\tAS:i:160\tXS:i:0\tXF:i:3\tXE:i:4\tXN:i:0\tRG:Z:g3
''')
out_fhand = StringIO()
standardize_sam(sam_fhand, out_fhand, 20, add_def_qual=True)
lines = out_fhand.getvalue().splitlines()
assert 'GGATGATNTTAGAG\t55555555555555\t' in lines[4]
assert lines[6].startswith('SGN-E40000\t20\t*\t0\t0\t*\t*\t0\t0\t')
assert lines[4].startswith('SGN-E200000\t0')
def run(self):
'''It runs the analysis.'''
self._log({'analysis_started':True})
settings = self._project_settings
project_path = settings['General_settings']['project_path']
tmp_dir = settings['General_settings']['tmpdir']
inputs = self._get_input_fpaths()
bam_paths = inputs['bams']
reference_path = inputs['reference']
output_dir = self._create_output_dirs()['result']
merged_bam_path = VersionedPath(os.path.join(output_dir,
BACKBONE_BASENAMES['merged_bam']))
merged_bam_fpath = merged_bam_path.next_version
#Do we have to add the default qualities to the sam file?
#do we have characters different from ACTGN?
add_qualities = settings['Sam_processing']['add_default_qualities']
#memory for the java programs
java_mem = settings['Other_settings']['java_memory']
picard_path = settings['Other_settings']['picard_path']
if add_qualities:
default_sanger_quality = settings['Other_settings']['default_sanger_quality']
default_sanger_quality = int(default_sanger_quality)
else:
default_sanger_quality = None
temp_dir = NamedTemporaryDir()
for bam_path in bam_paths:
bam_basename = bam_path.basename
temp_sam = NamedTemporaryFile(prefix='%s.' % bam_basename,
suffix='.sam')
sam_fpath = os.path.join(temp_dir.name, bam_basename + '.sam')
bam2sam(bam_path.last_version, temp_sam.name)
sam_fhand = open(sam_fpath, 'w')
# First we need to create the sam with added tags and headers
add_header_and_tags_to_sam(temp_sam, sam_fhand)
temp_sam.close()
sam_fhand.close()
#the standardization
temp_sam2 = NamedTemporaryFile(prefix='%s.' % bam_basename,
suffix='.sam', delete=False)
standardize_sam(open(sam_fhand.name), temp_sam2,
default_sanger_quality,
add_def_qual=add_qualities,
only_std_char=True)
temp_sam2.flush()
shutil.move(temp_sam2.name, sam_fhand.name)
temp_sam2.close()
get_sam_fpaths = lambda dir_: [os.path.join(dir_, fname) for fname in os.listdir(dir_) if fname.endswith('.sam')]
# Once the headers are ready we are going to merge
sams = get_sam_fpaths(temp_dir.name)
sams = [open(sam) for sam in sams]
temp_sam = NamedTemporaryFile(suffix='.sam')
reference_fhand = open(reference_path.last_version)
try:
merge_sam(sams, temp_sam, reference_fhand)
except Exception:
if os.path.exists(merged_bam_fpath):
os.remove(merged_bam_fpath)
raise
reference_fhand.close()
# close files
for sam in sams:
sam.close()
# Convert sam into a bam,(Temporary)
temp_bam = NamedTemporaryFile(suffix='.bam')
sam2bam(temp_sam.name, temp_bam.name)
# finally we need to order the bam
#print 'unsorted.bam', temp_bam.name
#raw_input()
sort_bam_sam(temp_bam.name, merged_bam_fpath,
java_conf={'java_memory':java_mem,
'picard_path':picard_path}, tmp_dir=tmp_dir )
temp_bam.close()
temp_sam.close()
create_bam_index(merged_bam_fpath)
self._log({'analysis_finished':True})