repeat proportion calculation in defuse

[fastq]

Hello,

How is the output field "repeat proportion" actually calculated, in detail?

thanks,
f

[Andrew]

Thanks for the question. For a given prediction, from a given cluster of read alignments, the alignment region on either side of the breakpoint is the region containing the start and end of all of the alignments on that side of the breakpoint. The repeat proportion is simply the proportion of the alignment region contained within a single repeat instance. If there are multiple overlapping repeats we take the max. There is some slight complication if the alignment region has multiple segments and is spread over the genome due to splicing, but we resolve this by just maximizing over all of the alignment segments.