naumenko-sa / crg Goto Github PK

Problem:
bcbio calls SVs with MetaSV on per sample basis, when analyzing families, we need to intersect individual SV reports to have one report for a family, something like bedtools multiint does.

Solution:

1. We are intersecting all SV intervals vs the indervals from the 1st sample (a random one).
2. For every SV event we report whether it is present in every sample.
3. If an event in one sample intersects with two events in the other sample, we report both events in the corresponding column.
4. We are NOT merging SV intervals, because of many potential bugs (DGV frequency in particular).
5. For the joint report we report DGV frequency for the event in the first sample and SVscores.
6. Final report format is CSV, use ; for multiple SV hits in one sample.