While refactoring the Pyrodigal code thanks to @jhahnfeld and @oschwengers, I noticed that the part of the score_connection function that is supposed to extract the frame with the maximum score was not actually extracting the maximum one while in training mode. This can change the predictions, for instance on GCF_000067165 the fixes causes a gene to start 160bp before what Prodigal actually predicts.
I'm not actually sure whether this is intended, but if it's not, here is a bugfix.
Hi again!
While refactoring the Pyrodigal code thanks to @jhahnfeld and @oschwengers, I noticed that the part of the
score_connectionfunction that is supposed to extract the frame with the maximum score was not actually extracting the maximum one while in training mode. This can change the predictions, for instance onGCF_000067165the fixes causes a gene to start 160bp before what Prodigal actually predicts.I'm not actually sure whether this is intended, but if it's not, here is a bugfix.