Currently, tallies of cleavages within a small "inner window" around each bp (default = 201bp, i.e. +/-100bp) slide by 1bp, with the large "outer window" sliding in lockstep, 1bp at a time. Adjacent "inner windows" of 201bp thus have 200bp of overlap; if a 201bp window has 4 cleavages near its right edge and there are no other cleavages within many bp, many successive observations of 4 will be processed, all of which go into the distribution of observations used to estimate the background. Highly overlapping windows contribute more data to the distribution than non-overlapping windows, which is a good thing, but the redundant/correlated contributions might be better replaced by quasi-independent observations from non-overlapping windows, although the latter will contribute much less data to the distribution. hotspot2 should allow for this.
Currently, tallies of cleavages within a small "inner window" around each bp (default = 201bp, i.e. +/-100bp) slide by 1bp, with the large "outer window" sliding in lockstep, 1bp at a time. Adjacent "inner windows" of 201bp thus have 200bp of overlap; if a 201bp window has 4 cleavages near its right edge and there are no other cleavages within many bp, many successive observations of 4 will be processed, all of which go into the distribution of observations used to estimate the background. Highly overlapping windows contribute more data to the distribution than non-overlapping windows, which is a good thing, but the redundant/correlated contributions might be better replaced by quasi-independent observations from non-overlapping windows, although the latter will contribute much less data to the distribution. hotspot2 should allow for this.