p. prediction for delins spanning two codons

[HLSZim]

In using normalizer, we ran into an issue with p. prediction/generation for two delins that involved more than one codon. In both cases, Normalizer normalized the c. as two dels separated by a single retained nt and the p. was generated as p.?, even though the correct predicted affected protein sequence is shown.

Examples:
https://mutalyzer.nl/normalizer/NM_000038.5:c.1997_1999delTACinsA
https://mutalyzer.nl/normalizer/NM_000038.5:c.497_499delCTAinsTT

Questions:
(1) is the normalization to two dels separated by a single retained nt intentional? I realize that it is arguably the correct HGVS nomenclature for delins that span multiple codons, although the SVD-WG has prepared a proposal to modify (SVD-EG010).
(2) if the normalization is intentional, is there a plan or timeline to adopt the SVD-WG proposal?
(3) is it intentional that p.? is generated even though the correct predicted affected protein sequence is shown?
(4) Is there another way to get a p. prediction for these types of variants?

Many thanks,
Heather

[jfjlaros]

Please find the answers to your questions below.

1. Indeed, this is expected behaviour.
2. The proposal has not been accepted yet, partly because of some fundamental concerns raised by our team. We are currently working on a revision of this proposal together with the SVD-WG / HVNC committee.
3. This is mainly because there are no rules to generate complex protein descriptions, such as the ones typically resulting from allele descriptions (compound variants). We do have a number of ideas, but we do not have any implementations yet.
4. There are multiple ways to get a description, e.g., by using Mutalyzer 2, which does not attempt to decompose and/or recombine an input description. This may give a satisfactory (maybe even correct) answer in this case, but in general, it will not.

[HLSZim]

Thanks for the rapid response!