Use a designated BGC with --query_bgc (not necessarily in your
--inputdir) and search only for similar (up to max(cutoffs)) BGCs in
your data set.
- TODO: after the first round of QBGC vs all, delete distance
information from all non-relevant distances