search

NBISweden / IgDiscover-legacy

Analyze antibody repertoires and discover new V genes from high-throughput sequencing reads
https://www.igdiscover.se
MIT License
17 stars 10 forks source link

germlinefilter: Discard cross-mapping artifacts even if they are whitelisted #57

Closed marcelm closed 7 years ago

marcelm commented 7 years ago

Database sequences that do not actually occur in the list of discovered candidate sequences get artificially re-introduced into the candidate list if they occur at a certain expression level. These are the ones marked with db in the cluster column.

The problem is that the expression level of such a sequence gets inflated if there is cross-mapping. Thus, the germline filtering rules should be updated such that even whitelisted/database sequences are subject to the cross-mapping correction step.

marcelm commented 7 years ago

This turns out to be a non-issue: Cross-mapping is actually only checked when the two involved sequences are database sequences.