Did you open the fasta file and check that position? Is there in fact a period there?
I'm not sure what this has to do with updating the version of qiime, do you mean that this import worked in an older version of qiime, but now doesn't work?
Yes. The fasta file has a "-" there and in many other places. We use it to show variable regions in a sequence or regions that may not have been fully characterised as part of a larger known sequence. This has worked fine on all the previous versions of qiime we have used. The issue has only arisen in running this command as part of preparing our classifiers for use in the newer version.
Sorry about the confusion of period or hyphen. We tried the hyphen first as that was what as in the fasta file originally but then we tried changing it to a period to see if that changed anything (it didn't). I must have copied across the second error message when we tried it with the period rather than the hyphen (the error messages were otherwise the same).
This was changed in 2021 to no longer allow periods or hyphens in the alphabet. I'm unsure of the reasoning around this decision. I will reach out to one of the developers who worked on this and get back to you.
For context, are you importing these sequences to later train a classifier on?
We noticed that your description of what you're using the hyphens to represent might be better represented as 'N', unless we're missing something. Hyphens generally indicate gaps in sequence alignments.