def main():
args = parse_args()
log_level = (logging.CRITICAL - 10 * args.v) if args.v else logging.INFO
log_file = args.log if args.log else 'compare_b2c.log'
logging.basicConfig(level=log_level, format='%(levelname)s - %(asctime)s %(message)s', filename=log_file)
if args.b and args.c:
bam_path = args.b
cram_path = args.c
if not utils.is_irods_path(bam_path) and not os.path.isfile(bam_path):
logging.error("This is not a file path: %s" % bam_path)
#sys.exit(1)
raise ValueError("This is not a file path: %s")
if not utils.is_irods_path(cram_path) and not os.path.isfile(cram_path):
logging.error("This is not a file path: %s" % cram_path)
#sys.exit(1)
raise ValueError("This is not a file path: %s")
errors = CompareStatsForFiles.compare_bam_and_cram_by_statistics(bam_path, cram_path)
if errors:
if args.e:
err_f = open(args.e, 'w')
for err in errors:
err_f.write(err + '\n')
err_f.close()
else:
print(errors)
sys.exit(1)
else:
logging.info("There were no errors and no differences between the stats for the 2 files.")
def main():
args = parse_args()
log_level = (logging.CRITICAL - 10 * args.v) if args.v else logging.INFO
log_file = args.log if args.log else 'compare_b2c.log'
logging.basicConfig(level=log_level,
format='%(levelname)s - %(asctime)s %(message)s',
filename=log_file)
if args.b and args.c:
bam_path = args.b
cram_path = args.c
#begin david
ref_path = ''
srun = ''
if args.r:
ref_path = args.r
if not os.path.isfile(ref_path):
logging.error("This is not a file path: %s" % ref_path)
raise ValueError("This is not a file path: %s")
if args.s:
srun = "srun"
#end david
if not utils.is_irods_path(bam_path) and not os.path.isfile(bam_path):
logging.error("This is not a file path: %s" % bam_path)
#sys.exit(1)
raise ValueError("This is not a file path: %s")
if not utils.is_irods_path(cram_path) and not os.path.isfile(
cram_path):
logging.error("This is not a file path: %s" % cram_path)
#sys.exit(1)
raise ValueError("This is not a file path: %s")
errors = CompareStatsForFiles.compare_bam_and_cram_by_statistics(
bam_path, cram_path, srun, ref_path)
if errors:
if args.e:
err_f = open(args.e, 'w')
for err in errors:
err_f.write(err + '\n')
err_f.close()
else:
print(errors)
sys.exit(1)
else:
logging.info(
"There were no errors and no differences between the stats for the 2 files."
)
def compare_bam_and_cram_by_statistics(cls, bam_path, cram_path):
errors = []
# Check that it's a valid file path
if not bam_path or (not utils.is_irods_path(bam_path) and not os.path.isfile(bam_path)):
errors.append("The BAM file path: %s is not valid" % bam_path)
if not cram_path or (not utils.is_irods_path(cram_path) and not os.path.isfile(cram_path)):
errors.append("The CRAM file path:%s is not valid" % cram_path)
if errors:
logging.error("There are errors with the file paths you provided: %s" % errors)
return errors
# Check that the files are readable by me
if not utils.is_irods_path(bam_path) and not utils.can_read_file(bam_path):
errors.append("Can't read file %s" % bam_path)
if not utils.is_irods_path(cram_path) and not utils.can_read_file(cram_path):
errors.append("Can't read file %s" % cram_path)
if errors:
logging.error("There are problems reading the files: %s" % errors)
return errors
# # Checking on samtools version:
# version_output = RunSamtoolsCommands.get_samtools_version_output()
# try:
# HandleSamtoolsVersion.check_samtools_version(version_output)
# except ValueError as e:
# errors.append(str(e))
# return errors
# Quickcheck the files before anything:
try:
RunSamtoolsCommands.run_samtools_quickcheck(bam_path)
except RuntimeError as e:
errors.append(str(e))
try:
RunSamtoolsCommands.run_samtools_quickcheck(cram_path)
except RuntimeError as e:
errors.append(str(e))
if errors:
logging.error("There are problems running quickcheck on the files you've given: %s" % errors)
return errors
# Calculate and compare flagstat:
try:
flagstat_b = RunSamtoolsCommands.get_samtools_flagstat_output(bam_path)
except RuntimeError as e:
errors.append(str(e))
try:
flagstat_c = RunSamtoolsCommands.get_samtools_flagstat_output(cram_path)
except RuntimeError as e:
errors.append(str(e))
if not errors:
errors.extend(cls.compare_flagstats(flagstat_b, flagstat_c))
else:
logging.error("THere are problems running flagstat on the files you've given: %s" % errors)
# Calculate and compare stats:
stats_fpath_b = bam_path + ".stats"
stats_fpath_c = cram_path + ".stats"
stats_b, stats_c = None, None
try:
stats_b = HandleSamtoolsStats.fetch_stats(bam_path, stats_fpath_b)
except (ValueError, RuntimeError) as e:
errors.append(str(e))
try:
stats_c = HandleSamtoolsStats.fetch_stats(cram_path, stats_fpath_c)
except (ValueError, RuntimeError) as e:
errors.append(str(e))
if not errors and stats_b and stats_c:
errors.extend(cls.compare_stats_by_sequence_checksum(stats_b, stats_c))
else:
errors.append("Can't compare samtools stats.")
logging.error("For some reason I can't compare samtools stats for your files.")
# Persist stats:
try:
if stats_b and not utils.is_irods_path(bam_path):
HandleSamtoolsStats.persist_stats(stats_b, stats_fpath_b)
except IOError as e:
errors.append("Can't save stats to disk for %s file" % bam_path)
logging.error("Can't save stats to disk for %s file" % bam_path)
try:
if stats_c and not utils.is_irods_path(cram_path):
HandleSamtoolsStats.persist_stats(stats_c, stats_fpath_c)
except IOError as e:
errors.append("Can't save stats to disk for %s file" % cram_path)
logging.error("Can't save stats to disk for %s file" % cram_path)
return errors
def compare_bam_and_cram_by_statistics(cls, bam_path, cram_path):
errors = []
# Check that it's a valid file path
if not bam_path or (not utils.is_irods_path(bam_path)
and not os.path.isfile(bam_path)):
errors.append("The BAM file path: %s is not valid" % bam_path)
if not cram_path or (not utils.is_irods_path(cram_path)
and not os.path.isfile(cram_path)):
errors.append("The CRAM file path:%s is not valid" % cram_path)
if errors:
logging.error(
"There are errors with the file paths you provided: %s" %
errors)
return errors
# Check that the files are readable by me
if not utils.is_irods_path(bam_path) and not utils.can_read_file(
bam_path):
errors.append("Can't read file %s" % bam_path)
if not utils.is_irods_path(cram_path) and not utils.can_read_file(
cram_path):
errors.append("Can't read file %s" % cram_path)
if errors:
logging.error("There are problems reading the files: %s" % errors)
return errors
# # Checking on samtools version:
# version_output = RunSamtoolsCommands.get_samtools_version_output()
# try:
# HandleSamtoolsVersion.check_samtools_version(version_output)
# except ValueError as e:
# errors.append(str(e))
# return errors
# Quickcheck the files before anything:
try:
RunSamtoolsCommands.run_samtools_quickcheck(bam_path)
except RuntimeError as e:
errors.append(str(e))
try:
RunSamtoolsCommands.run_samtools_quickcheck(cram_path)
except RuntimeError as e:
errors.append(str(e))
if errors:
logging.error(
"There are problems running quickcheck on the files you've given: %s"
% errors)
return errors
# Calculate and compare flagstat:
try:
flagstat_b = RunSamtoolsCommands.get_samtools_flagstat_output(
bam_path)
except RuntimeError as e:
errors.append(str(e))
try:
flagstat_c = RunSamtoolsCommands.get_samtools_flagstat_output(
cram_path)
except RuntimeError as e:
errors.append(str(e))
if not errors:
errors.extend(cls.compare_flagstats(flagstat_b, flagstat_c))
else:
logging.error(
"THere are problems running flagstat on the files you've given: %s"
% errors)
# Calculate and compare stats:
stats_fpath_b = bam_path + ".stats"
stats_fpath_c = cram_path + ".stats"
stats_b, stats_c = None, None
try:
stats_b = HandleSamtoolsStats.fetch_stats(bam_path, stats_fpath_b)
except (ValueError, RuntimeError) as e:
errors.append(str(e))
try:
stats_c = HandleSamtoolsStats.fetch_stats(cram_path, stats_fpath_c)
except (ValueError, RuntimeError) as e:
errors.append(str(e))
if not errors and stats_b and stats_c:
errors.extend(
cls.compare_stats_by_sequence_checksum(stats_b, stats_c))
else:
errors.append("Can't compare samtools stats.")
logging.error(
"For some reason I can't compare samtools stats for your files."
)
# Persist stats:
try:
if stats_b and not utils.is_irods_path(bam_path):
HandleSamtoolsStats.persist_stats(stats_b, stats_fpath_b)
except IOError as e:
errors.append("Can't save stats to disk for %s file" % bam_path)
logging.error("Can't save stats to disk for %s file" % bam_path)
try:
if stats_c and not utils.is_irods_path(cram_path):
HandleSamtoolsStats.persist_stats(stats_c, stats_fpath_c)
except IOError as e:
errors.append("Can't save stats to disk for %s file" % cram_path)
logging.error("Can't save stats to disk for %s file" % cram_path)
return errors
