Integration with tidytcells about stitchr HOT 4 CLOSED

Hi Yuta,

Thanks, very nice of you to say!

I had indeed seen tidytcells, which looks great. I'm open to seeing how this could integrate - maybe I'll reach out to discuss how best this could work. Either way I'm very happy pointing that out in the stitchr docs, as I've definitely had people hit stitchr errors due to formatting errors in their input allele names!

from stitchr.

Amazing!

In terms of integration I had just thought that when the input species was either human or mouse, that tidytcells could check for IMGT compliance automatically, and either attempt auto-correction, or notify the user if the input symbol is not recoverable. But I also appreciate that having a dependency on a small project like tidytcells is a point of failure which you might feel weary of.

In either case, maybe a mention in the docs could potentially help out some future users with messy data that they want to put through stitchr. 💪🏼 If you're open to it, I can make some time in the near future to add a tiny mention in the docs and submit a pull request, if that makes life easier for you?

from stitchr.

I'm actually working on some minor tweaks at the moment, including the docs, so I can add in the suggestion no problem.

The only thing that makes me a little nervous about instituting an automatic check is that it might not cope so well with odd/uncommon/non-natural sequences (which are things that I often put through myself, and which did upset things in earlier test versions). For example I've recently been playing with some funny ORF alleles, like the human TRBV7-3*03, which uses Arg instead of the usual conserved second Cys, which I think tidytcells might break the sequences of. Hence generally I've mostly made stitchr to work under the assumption that users are best placed to get their own data sorted upstream, and just instead raise warnings if it thinks something might have gone awry.

However I can certainly imagine that many users won't care about those edge cases, especially when trying to adapt large datasets. I think maybe it could be included as an optional check, triggered via a flag when running. Perhaps even separate flags for tidytcells.tr.standardize and tidytcells.junction.standardize, to allow users a bit more flexibility.

from stitchr.

Thanks for adding a note in the docs! I don't know if my email reply to you came through but I see what you mean about edge cases and that perhaps best practice is to have standardisation and stitching be two separate steps. Anyways I hope our tools combined can have a synergistic effect and help some TCR people out there :).

from stitchr.