[URGENT] Use BLAST start/stop data to understand whether two ORFs are a duplication or fragments

[lucventurini]

Currently, the BLAST algorithm is quite stupid ... we only look whether two ORFs share a hit, without considering whether they hit the same region of the target. This means that two duplicated genes, fused by one method, will end up being considered as fragments of the same gene and therefore the ORFs will not get splitted.

This should be easily fixable by considering whether the hits shared by two ORFs are or are not overlapping. Target: "check_split_by_blast" in Transcript.

[lucventurini]

At the moment, this feature can be controlled using the following parameter (float, percentage between 0 and 1) in the "chimera_split" section:

• "min_overlap_duplication" (default 90%)

This means that if the overlap between HSPs on two different ORFs have an overlap that is greater than 90% of the length of the target, the hit should be considered a tandem duplication and therefore split.