update hitaggregator function

[ccrobertson]

Here is a modified version of the hit_aggregator() function that does the following:

(1) takes all peak regions from all narrowPeak files in the project and defines a set of k non-overlapping regions using GenomicRanges “reduce”
(2) assigns each observed peak across narrowPeak files to one of the k non-overlapping regions defined in (1)
(3) for each of the k non-overlapping regions, it finds which observed peak has the highest score

The function returns a list of k regions with peak boundaries based on the highest scoring peaks in step (3).

[ccrobertson]

I'm not sure my modification is actually what you want "hit_aggregator" to do. This was my attempt to ensure that the resulting consensus peak set is non-overlapping (currently I find that sometimes there is some overlap between peaks)... An alternative approach could be to apply "reduce" to the results from the original implementation at the end, instead of defining a disjoint set up front.