Peaks are splitted into smaller peaks if they are longer than target_size. The splitting is done by first checking
if the peak has a length above very_long. If it does, the peak is splitted arbitrarily into chunks of size that is approximately very_long.
Then, the chunks are splitted into smaller chunks of approximately size target_size. This splitting is done by first finding the
best offset to start from and then splitting the peaks into chunks of size target_size starting from that offset.
Detection of the best offset is done by substracting the mean of each interval from each coverage, removing values which became
zero, and then correlating thr marginal coverage with simulated 'triangle' peaks starting at different offsets.
Usage
split_long_peaks(
marginal_track,
peaks,
target_size = 500,
max_peak_size = 1000,
very_long = 5000,
min_peak_size = NULL,
window_size = target_size
)Arguments
- marginal_track
Name of the 'misha' track with the marginal coverage. You can create it using
import_atac_marginal.- peaks
An intervals set with the peaks to split.
- target_size
The target size of peaks.
- max_peak_size
Peaks above this size would be splitted into smaller peaks.
- very_long
Peaks above this size would be splitted arbitrarily into smaller peaks before fitting the best offset.
- min_peak_size
Peaks below this size would be discarded.
- window_size
Smoothing window size (from each side of the iterator) to use of the marginal track. Default: traget_size.