def load_scores_wiggle(fname, chrom_buffer_size=3):
"""
Read a wiggle file and return a dict of BinnedArray objects keyed
by chromosome.
"""
scores_by_chrom = dict()
try:
for chrom, pos, val in bx.wiggle.Reader(UCSCOutWrapper(open(fname))):
if chrom not in scores_by_chrom:
if chrom_buffer_size:
scores_by_chrom[chrom] = BinnedArray()
chrom_buffer_size -= 1
else:
scores_by_chrom[chrom] = PositionalScoresOnDisk()
scores_by_chrom[chrom][pos] = val
except UCSCLimitException:
# Wiggle data was truncated, at the very least need to warn the user.
print(
'Encountered message from UCSC: "Reached output limit of 100000 data values", so be aware your data was truncated.'
)
except IndexError:
stop_err(
'Data error: one or more column data values is missing in "%s"' %
fname)
except ValueError:
stop_err(
'Data error: invalid data type for one or more values in "%s".' %
fname)
return scores_by_chrom
def main():
if len(sys.argv) > 1:
in_file = open(sys.argv[1])
else:
in_file = open(sys.stdin)
if len(sys.argv) > 2:
out_file = open(sys.argv[2], "w")
else:
out_file = sys.stdout
try:
for fields in bx.wiggle.IntervalReader(UCSCOutWrapper(in_file)):
out_file.write("%s\n" % "\t".join(map(str, fields)))
except UCSCLimitException:
# Wiggle data was truncated, at the very least need to warn the user.
print(
'Encountered message from UCSC: "Reached output limit of 100000 data values", so be aware your data was truncated.'
)
except ValueError as e:
in_file.close()
out_file.close()
stop_err(str(e))
in_file.close()
out_file.close()
def main():
with open(sys.argv[1]) as in_file, open(sys.argv[2], "w") as out_file:
try:
for fields in bx.wiggle.IntervalReader(UCSCOutWrapper(in_file)):
out_file.write("%s\n" % "\t".join(map(str, fields)))
except UCSCLimitException:
# Wiggle data was truncated, at the very least need to warn the user.
sys.stderr.write(
'Encountered message from UCSC: "Reached output limit of 100000 data values", so be aware your data was truncated.'
)