I’ve noticed that using IgBlast/Changeo, I’m getting sequences annotated as functional and in-frame, which do not produce functional antibodies.
These sequences have indels in the J region FW4, so that while the V and the start of the J are in-frame, the bulk of the J is not in-frame, and the constant region will not be in-frame either.
I’m assigning these sequences are non-productive for the purposes of, “If you add a constant region vector on the end of the recovered VDJ sequence, does it make a protein?”.
I realise this is subtlety different to the in-frame junction and contains no stop codons definition used by igblast/changeo. How are other people dealing with indels in antibody sequences for classification?