Default lambda model has implausible J deletions

[psathyrella]

It seems like the default lambda model has J 5' deletion lengths for some J genes that don't seem very plausible, for instance at line 328 (IGLJ5*01 I think):

https://github.com/statbiophys/OLGA/blob/master/olga/default_models/human_B_lambda/model_marginals.txt#L328

~all the probability is at length 21, which would delete beyond the conserved J codon. Lines 334 (all prob at len=21) and 336 (all probs 0) also seem problematic:

These issues seem to result in zero probability for any realistic CDR3 that I can pass in, which also leads to zero for any lambda passed into Sonnia.

[giulioisac]

Hello, that's a good point. We inferred this recently on a new dataset. I'll try to fix it, in the meantime you can check out the old model,. It might be enough for your needs.