def mark_duplicates(bam_path, output_bam_path, output_metrics_path):
cmd = "java -Xmx1g -jar {} \
INPUT={} \
OUTPUT={} \
METRICS_FILE={}"\
.format(_MARK_DUPLICATES, bam_path, output_bam_path, output_metrics_path)
common.system(cmd)
common.assert_file(output_bam_path, cmd)
common.assert_file(output_metrics_path, cmd)
return output_bam_path
def do_errors(args):
"""Compare the existence of output.gd files with that of *.log.txt, if there is not an
output.gd for for a log file then it must have failed and needs to be brought to the
users attention."""
output_search = os.path.join(args.output_dir, "*/output/output.gd")
log_search = os.path.join(args.log_dir,"*.log.txt")
print >> sys.stderr, "GenomeDiff search:", output_search
print >> sys.stderr, "Log search:", log_search
output_paths = glob.glob(output_search)
assert output_paths, "No output/output.gd's found for search: " + output_search
output_paths = (path.replace("/output/output.gd", "") for path in output_paths)
output_paths = map(os.path.basename, output_paths)
output_names = set(output_paths)
log_paths = glob.glob(log_search)
assert log_paths, "No *.log.txt's found for search: " + log_search
log_paths = map(os.path.basename, log_paths)
log_paths = (path.replace(".log.txt", "") for path in log_paths)
log_names = set(log_paths)
error_names = log_names.difference(output_names)
if not len(error_names):
print >> sys.stderr, "No errors!"
return 0
error_paths = [os.path.join(args.log_dir, name) + ".log.txt" for name in error_names]
for path in error_paths:
print 20 * '*'
common.system("tail " + path)
print 20 * '*'
print
return -1
def view(bam_path, sam_path):
cmd = samtools_exe + " view -hbS -o {} {}".format(bam_path, sam_path)
common.system(cmd)
common.assert_file(bam_path, cmd)
return bam_path
def faidx(fasta_path):
cmd = samtools_exe + " faidx {}".format(fasta_path)
common.system(cmd)
return fasta_path
def mpileup(fasta_path, bam_sorted_path, out_vcf_path):
cmd = samtools_exe + " mpileup -uf {} {} | bcftools view -vcg - > {}".format(fasta_path, bam_sorted_path, out_vcf_path)
common.system(cmd)
common.assert_file(out_vcf_path, cmd)
return out_vcf_path
def index(bam_path):
cmd = samtools_exe + " index {}".format(bam_path)
common.system(cmd)
common.assert_file(bam_path, cmd)
return bam_path
def merge(sam_paths, merged_bam_path, bam_paths):
cmd = samtools_exe + " merge -r -n -h {} {} {}".format(" -h ".join(sam_paths), merged_bam_path, " ".join(bam_paths))
common.system(cmd)
common.assert_file(merged_bam_path, cmd)
return merged_bam_path
def sort(bam_path, bam_sorted_prefix):
cmd = samtools_exe + " sort {} {}".format(bam_path, bam_sorted_prefix)
sorted_bam_path = bam_sorted_prefix + ".bam"
common.system(cmd)
common.assert_file(sorted_bam_path, cmd)
return sorted_bam_path
def merge_sams(sam_paths, output, merge_seq_dicts = True):
cmd = "java -jar {} INPUT={} OUTPUT={} MERGE_SEQUENCE_DICTIONARIES={}".format(_MERGE_SAMS," INPUT=".join(sam_paths), output, "true" if merge_seq_dicts else "false")
common.system(cmd)
common.assert_file(output, cmd)
return output
def sort_sam(aln_path, output_aln_path, sort_option = "coordinate"):
cmd = "java -jar {} INPUT={} OUTPUT={} SORT_ORDER={}".format(_SORT_SAM, aln_path, output_aln_path, sort_option)
common.system(cmd)
common.assert_file(output_aln_path, cmd)
return output_aln_path
def create_sequence_dictionary(ref_path, output_path):
cmd = "java -jar {} R={} O={}".format(_CREATE_SEQUENCE_DICTIONARY, ref_path, output_path)
common.system(cmd)
common.assert_file(output_path, cmd)
return output_path
def add_or_replace_read_groups(input_path, output_path):
cmd = "java -jar {} I={} O={} LB=FOO PL=ILLUMINA PU=BAR SM=NEE".format(_ADD_OR_REPLACE_READ_GROUPS, input_path, output_path)
common.system(cmd)
common.assert_file(output_path, cmd)
return output_path
def validate_alignment(aln_path):
cmd = "java -jar {} I={}".format(_VALIDATE_SAM_FILE, aln_path)
common.system(cmd)
return aln_path