We just had a case internally of a blacklist containing duplicate and neighboring entries. Due to how the blacklisting stuff works, this resulted in something like 150% of the reads being estimated as falling in a blacklist region...which is obviously not possible. What we need to do is have a step after reading in the blacklist were we go through the list of regions and ensure that (A) the don't overlap and (B) they aren't next to each other. Ideally the regions shouldn't be within a few hundred bases of each other either.