def slice_pooled_peaks(self,
threshold,
pooled_threshold,
rep_files,
pseudorep_files,
pooled_files,
pooled_peaks,
output_dir,
ranking_measure='tag-count'):
idrutil = IdrUtilities()
# Determine how many peaks we want to keep.
keep_count = idrutil.get_peaks_within_threshold(threshold, rep_files)
idrutil.get_peaks_within_threshold(threshold, pseudorep_files)
pooled_count = idrutil.get_peaks_within_threshold(
pooled_threshold, pooled_files)
# Pooled count should be within 2-fold of keep_count
if abs(math.log(keep_count / pooled_count, 2)) > 1:
print('!! Warning: The number of peaks within the replicate ' +
'threshold is not within two-fold of the number of ' +
'peaks within the pooled threshold. This could indicate ' +
'inconsistencies in the datasets.\n' +
'Replicate count: {}, Pooled count: {}'.format(
keep_count, pooled_count))
# Slice our pooled peak file accordingly.
output_file = idrutil.slice_peaks(pooled_peaks, keep_count,
ranking_measure, output_dir)
print('{} peaks output to {}'.format(keep_count, output_file))
def slice_pooled_peaks(
self,
threshold,
pooled_threshold,
rep_files,
pseudorep_files,
pooled_files,
pooled_peaks,
output_dir,
ranking_measure="tag-count",
):
idrutil = IdrUtilities()
# Determine how many peaks we want to keep.
keep_count = idrutil.get_peaks_within_threshold(threshold, rep_files)
idrutil.get_peaks_within_threshold(threshold, pseudorep_files)
pooled_count = idrutil.get_peaks_within_threshold(pooled_threshold, pooled_files)
# Pooled count should be within 2-fold of keep_count
if abs(math.log(keep_count / pooled_count, 2)) > 1:
print(
"!! Warning: The number of peaks within the replicate "
+ "threshold is not within two-fold of the number of "
+ "peaks within the pooled threshold. This could indicate "
+ "inconsistencies in the datasets.\n"
+ "Replicate count: {}, Pooled count: {}".format(keep_count, pooled_count)
)
# Slice our pooled peak file accordingly.
output_file = idrutil.slice_peaks(pooled_peaks, keep_count, ranking_measure, output_dir)
print("{} peaks output to {}".format(keep_count, output_file))
