pick a valid allele for serotypes

[timodonnell]

Calling normalize_allele_name on serotypes A24, B07, or B44 currently gives the invalid allele names HLA-A*24:01, HLA-B*07:01, and HLA-B*44:01, respectively.

To do this correctly we should consult a list of valid allele names, or add a function that maps serotypes to a list of alleles.

[rleonid]

To the best of my knowledge, a mapping is found in the HLA dictionary, whose most recent version is from 2008. As you know this is not a bijection, and many alleles can have the same serotype. To avoid the pdf, IMGT's web interface conveniently provides the "expert"[1] and WHO assigned serotypes down in the HLA dictionary data section: https://www.ebi.ac.uk/cgi-bin/ipd/imgt/hla/get_allele.cgi?A*01:01:01:01
There is some concordance between the two.

1. Presumably, the WHO also has some experts, but who am I to judge?

[iskandr]

This seems related to #8 in that an input string like "HLA-A class I" should return a large set of alleles, whereas serotypes should return a more narrow set.

[iskandr]

One idea for an API is to have the most general parse function return a list of candidates.