The problem I am having is that there can be several transcripts per gene, so when the the variant is described as "BRCA1, ARG1203TER", it seems ambiguous to me which transcript it is referring to. Does anybody know how to resolve this ambiguity?
I'm not sure if you can. I may be wrong, but I think OMIM reports allelic variants as they were reported in the original publication. Which can be a problem given multiple transcripts, updates to sequences and gene models, etc. I would hazard that in most recent cases they will match up with the RefSeq numbering, but I doubt you can guarantee that. I don't know how much overlap there is between the two but you could also check out HUMSAVAR and see if their data is easier to resolve absolutely. It should be based on Uniprot reference sequences.