keithchev
commented
1 month ago It feels fine and even good to me that the final output is opinionated (that is, filtered). My general intuition is that most users would want less rather than more information (i.e. just a list of "true" bioactive peptides) and wouldn't care about the various categories of things that are filtered out of the final list.
A kind of compromise could be to add a note somewhere to make extra sure that the user understands that this filtering is happening and to explain where the propeptides can be found in the intermediate output. That would certainly feel more than fair/transparent to me.
DeepPeptide predicts cleavage peptides from amino acid sequences. It reports both peptides and propeptides. For propeptides, it uses the uniprot definition: "a part of a protein that is cleaved during maturation or activation. Once cleaved, a propeptide generally has no independent biological function."
I keep removing these because they aren't bioactive peptides, which is what peptigate is designed to detect. I keep leaning toward keeping them in because it's more information that a user could look at, but I think they're actually pretty confusing. Because we've "advertised" peptigate as a tool that predicts bioactive peptides, one might not expect that the largest class of peptides predicted by deeppeptide would actually be non-bioactive peptides. If we do filter them, they would still be reported in the deeppeptide intermediate output, so a user could find them if they wanted to.
@keithchev would appreciate input if you have a second!