Open josephwb opened 3 years ago
@josephwb I feel like this is something we talked about like 5 years ago and decided that it didn't need to be in multiples of three, but now I pretty strongly feel like it should. With all the automated pipelines that exists, the more checks the better.
Thank you @jfwalker. I will put an enforced check. If necessary, I can put in a -f --force
do-it-anyway option for both.
Should also consider removing terminal (stop) codons here, as they can be a pain.
+1 for "remove extra junk" option. I would love that feature. I am running into cases were "nucleotide alignment involves 191 codons, but protein alignment involves 190 amino acids. Skipping" In my cases, it is a leading codon that is absent in the AA alignment. I would love if it was possible to ignore the missing leading codon and output only the 190 codons that correspond to the 190 AA.
Found investigating #143.
pxaa2cdn
won't work if nucleotide sequence lengths are not a multiple of 3 (codons). However,pxtlate
will translate as much as it can, and just disregard any "extra" nucleotides. Basically, I was unable to construct an example dataset forpxaa2cnd
by usingpxtlate
, which is worrying,Seems like both programs should share a policy: either enforce multiples of 3 in a draconian manner, or be chill and let what ever get passed through.
I like the "force the user to use good data" route, but do not do this enough to know how often nuc seqs will have "extra" nucleotides. And from working on the
pxaa2cdn
example I know how much of a pain it can be to wrangle seq lengths.I guess @jfwalker would be the best to decide which policy above is most appropriate.