def conservation_serial(bed_file, window_sizes, chrom_bigwig_dict):
# output header fields
fields = ['name', 'position', 'transcript_length', 'mean']
fields.extend(map(str, window_sizes))
print '\t'.join(fields)
# process bed file
for f in BEDFeature.parse(open(bed_file)):
# retrieve conservation data
bigwig_file = chrom_bigwig_dict[f.chrom]
arr = extract_bigwig_data(f, bigwig_file)
# measure conservation at various sliding windows
window_scores = []
for window_size in window_sizes:
window_scores.append(best_sliding_window(arr, window_size,
np.mean))
# calc mean conservation
finitearr = arr[np.isfinite(arr)]
if len(finitearr) == 0:
mean_cons = np.nan
else:
mean_cons = np.mean(finitearr)
fields = [
f.name,
'%s:%d-%d[%s]' % (f.chrom, f.tx_start, f.tx_end, f.strand),
str(len(arr)),
str(mean_cons)
]
fields.extend(map(str, window_scores))
print '\t'.join(fields)
def _consumer(input_queue, output_queue):
while True:
line = input_queue.get()
if line is None:
break
f = BEDFeature.from_string(line)
# retrieve conservation data
bigwig_file = chrom_bigwig_dict[f.chrom]
arr = extract_bigwig_data(f, bigwig_file)
# measure conservation at various sliding windows
window_scores = []
for window_size in window_sizes:
window_scores.append(best_sliding_window(arr, window_size, np.mean))
# measure average conservation
finitearr = arr[np.isfinite(arr)]
if len(finitearr) == 0:
mean_cons = np.nan
else:
mean_cons = np.mean(finitearr)
fields = [f.name, '%s:%d-%d[%s]' % (f.chrom, f.tx_start, f.tx_end, f.strand),
str(len(arr)), str(mean_cons)]
fields.extend(map(str,window_scores))
result = '\t'.join(fields)
output_queue.put(result)
output_queue.put(None)
def _consumer(input_queue, output_queue):
while True:
line = input_queue.get()
if line is None:
break
f = BEDFeature.from_string(line)
# retrieve conservation data
bigwig_file = chrom_bigwig_dict[f.chrom]
arr = extract_bigwig_data(f, bigwig_file)
# measure conservation at various sliding windows
window_scores = []
for window_size in window_sizes:
window_scores.append(
best_sliding_window(arr, window_size, np.mean))
# measure average conservation
finitearr = arr[np.isfinite(arr)]
if len(finitearr) == 0:
mean_cons = np.nan
else:
mean_cons = np.mean(finitearr)
fields = [
f.name,
'%s:%d-%d[%s]' % (f.chrom, f.tx_start, f.tx_end, f.strand),
str(len(arr)),
str(mean_cons)
]
fields.extend(map(str, window_scores))
result = '\t'.join(fields)
output_queue.put(result)
output_queue.put(None)
def conservation_serial(bed_file, chrom_bigwig_dict, sig_threshold,
window_sizes, num_processes):
# process bed file
for f in BEDFeature.parse(open(bed_file)):
fields = feature_conservation(f, chrom_bigwig_dict, sig_threshold,
window_sizes)
print '\t'.join(fields)
def conservation_serial(bed_file, chrom_bigwig_dict, sig_threshold,
window_sizes, num_processes):
# process bed file
for f in BEDFeature.parse(open(bed_file)):
fields = feature_conservation(f, chrom_bigwig_dict,
sig_threshold, window_sizes)
print '\t'.join(fields)
def _consumer(input_queue, output_queue):
while True:
line = input_queue.get()
if line is None:
break
f = BEDFeature.from_string(line)
fields = feature_conservation(f, chrom_bigwig_dict, sig_threshold,
window_sizes)
output_queue.put('\t'.join(fields))
output_queue.put(None)
def _consumer(input_queue, output_queue):
while True:
line = input_queue.get()
if line is None:
break
f = BEDFeature.from_string(line)
fields = feature_conservation(f, chrom_bigwig_dict,
sig_threshold, window_sizes)
output_queue.put('\t'.join(fields))
output_queue.put(None)
def main():
logging.basicConfig(level=logging.DEBUG,
format="%(asctime)s - %(name)s - %(levelname)s - %(message)s")
parser = argparse.ArgumentParser()
parser.add_argument('bed_file')
args = parser.parse_args()
trees = collections.defaultdict(lambda: ClusterTree(0,1))
for f in BEDFeature.parse(open(args.bed_file)):
tree = trees[f.chrom]
for start,end in f.exons:
tree.insert(start, end, 1)
footprint = 0
for chrom in sorted(trees):
chromprint = 0
tree = trees[chrom]
for start,end,indexes in tree.getregions():
chromprint += (end - start)
#print chrom, chromprint
footprint += chromprint
print 'total', footprint
def main():
logging.basicConfig(
level=logging.DEBUG,
format="%(asctime)s - %(name)s - %(levelname)s - %(message)s")
parser = argparse.ArgumentParser()
parser.add_argument('bed_file')
args = parser.parse_args()
trees = collections.defaultdict(lambda: ClusterTree(0, 1))
for f in BEDFeature.parse(open(args.bed_file)):
tree = trees[f.chrom]
for start, end in f.exons:
tree.insert(start, end, 1)
footprint = 0
for chrom in sorted(trees):
chromprint = 0
tree = trees[chrom]
for start, end, indexes in tree.getregions():
chromprint += (end - start)
#print chrom, chromprint
footprint += chromprint
print 'total', footprint
def conservation_serial(bed_file, window_sizes, chrom_bigwig_dict):
# output header fields
fields = ['name', 'position', 'transcript_length', 'mean']
fields.extend(map(str,window_sizes))
print '\t'.join(fields)
# process bed file
for f in BEDFeature.parse(open(bed_file)):
# retrieve conservation data
bigwig_file = chrom_bigwig_dict[f.chrom]
arr = extract_bigwig_data(f, bigwig_file)
# measure conservation at various sliding windows
window_scores = []
for window_size in window_sizes:
window_scores.append(best_sliding_window(arr, window_size, np.mean))
# calc mean conservation
finitearr = arr[np.isfinite(arr)]
if len(finitearr) == 0:
mean_cons = np.nan
else:
mean_cons = np.mean(finitearr)
fields = [f.name, '%s:%d-%d[%s]' % (f.chrom, f.tx_start, f.tx_end, f.strand),
str(len(arr)), str(mean_cons)]
fields.extend(map(str,window_scores))
print '\t'.join(fields)
def main():
logging.basicConfig(
level=logging.DEBUG,
format="%(asctime)s - %(name)s - %(levelname)s - %(message)s")
parser = argparse.ArgumentParser()
parser.add_argument('-n', type=int, dest='num_samples', default=100000)
parser.add_argument('genome_bed_file')
parser.add_argument('region_lengths_file')
# parse arguments
args = parser.parse_args()
genome_bed_file = args.genome_bed_file
region_lengths_file = args.region_lengths_file
num_samples = max(1, args.num_samples)
# check arguments
if not os.path.exists(genome_bed_file):
parser.error('genome bed file %s not found' % (genome_bed_file))
if not os.path.exists(region_lengths_file):
parser.error('region lengths file %s not found' %
(region_lengths_file))
# get transcript lengths from transcript bed file
logging.debug('Reading region lengths')
region_lengths = []
with open(region_lengths_file) as f:
region_lengths.extend(int(x.strip()) for x in f)
# first pass to index windows
logging.debug('Reading genome bed file')
features = []
genome_cumsums = []
genome_size = 0
max_feature_size = 0
for f in BEDFeature.parse(open(genome_bed_file)):
length = sum((e[1] - e[0]) for e in f.exons)
if length == 0:
continue
features.append(f)
genome_cumsums.append(genome_size)
genome_size += length
max_feature_size = max(max_feature_size, length)
logging.debug('Genome bed size %d' % (genome_size))
# get windows
trials = 0
windows = []
for i in xrange(num_samples):
# choose random region length
length = region_lengths[random.randrange(len(region_lengths))]
while True:
trials += 1
# choose random window start
genome_start = random.randrange(genome_size - length)
feature_index = bisect.bisect_right(genome_cumsums,
genome_start) - 1
f = features[feature_index]
feature_length = (f.tx_end - f.tx_start)
# can use available length of this feature
genome_feature_start = genome_cumsums[feature_index]
feature_offset = genome_start - genome_feature_start
avail_length = feature_length - feature_offset
if avail_length < length:
continue
windows.append((f.chrom, f.tx_start + feature_offset,
f.tx_start + feature_offset + length))
break
if (i > 0) and (i % 1000) == 0:
logging.debug('Finished %d' % (i))
logging.debug('Sampled %d windows (%d trials)' % (len(windows), trials))
for window in sorted(windows):
print '\t'.join(map(str, window))
return 0
def main():
logging.basicConfig(level=logging.DEBUG,
format="%(asctime)s - %(name)s - %(levelname)s - %(message)s")
parser = argparse.ArgumentParser()
parser.add_argument('-n', type=int, dest='num_samples', default=100000)
parser.add_argument('genome_bed_file')
parser.add_argument('region_lengths_file')
# parse arguments
args = parser.parse_args()
genome_bed_file = args.genome_bed_file
region_lengths_file = args.region_lengths_file
num_samples = max(1, args.num_samples)
# check arguments
if not os.path.exists(genome_bed_file):
parser.error('genome bed file %s not found' % (genome_bed_file))
if not os.path.exists(region_lengths_file):
parser.error('region lengths file %s not found' % (region_lengths_file))
# get transcript lengths from transcript bed file
logging.debug('Reading region lengths')
region_lengths = []
with open(region_lengths_file) as f:
region_lengths.extend(int(x.strip()) for x in f)
# first pass to index windows
logging.debug('Reading genome bed file')
features = []
genome_cumsums = []
genome_size = 0
max_feature_size = 0
for f in BEDFeature.parse(open(genome_bed_file)):
length = sum((e[1]-e[0]) for e in f.exons)
if length == 0:
continue
features.append(f)
genome_cumsums.append(genome_size)
genome_size += length
max_feature_size = max(max_feature_size, length)
logging.debug('Genome bed size %d' % (genome_size))
# get windows
trials = 0
windows = []
for i in xrange(num_samples):
# choose random region length
length = region_lengths[random.randrange(len(region_lengths))]
while True:
trials += 1
# choose random window start
genome_start = random.randrange(genome_size - length)
feature_index = bisect.bisect_right(genome_cumsums, genome_start) - 1
f = features[feature_index]
feature_length = (f.tx_end - f.tx_start)
# can use available length of this feature
genome_feature_start = genome_cumsums[feature_index]
feature_offset = genome_start - genome_feature_start
avail_length = feature_length - feature_offset
if avail_length < length:
continue
windows.append((f.chrom, f.tx_start + feature_offset, f.tx_start + feature_offset + length))
break
if (i > 0) and (i % 1000) == 0:
logging.debug('Finished %d' % (i))
logging.debug('Sampled %d windows (%d trials)' % (len(windows), trials))
for window in sorted(windows):
print '\t'.join(map(str, window))
return 0
