Incorrect DPB1 TCE3 match grading

[RimkePostel]

Issue DPB1 permissiveness is currently shown incorrectly. Donors with a DPB1*04:01:01G or DPB1*01:01:01G typing are scored as ambiguous when they should be scored as non-permissive (in either GvH or HvG direction) or permissive. We were experiencing similar issues with Hap-E. The Hap-E developers found that this issue resulted from the new HLA release 3.53.0 where several rare alleles got a new TCE assignment.

To Reproduce Steps to reproduce the behavior:

1. Go to Search&Match
2. Add a patient with DPB1*01:01:01 and DPB1*04:01:01 alleles
3. Several donors will be returned as ambiguous when they should be returned as permissive or non-permissive

Expected behaviour We expect the DPB1 match grade permissiveness to match with the permissiveness that is calculated by the IMGT/HLA DPB1 T-Cell Epitope Algorithm.

Screenshots

Additional context For example case, see: search id = d6304d8b-e880-4ca0-ac5c-2ae67df94129 donor ids: 377449095 377458494

[RimkePostel]

Hi @zabeen ,

As requested I am sending you additional information about how the Hap-E developers have fixed this issue:

Since the new HLA release 3.53.0 the allele DPB1*1148:01 of G group DPB1*04:01:01G and DPB1*1038:01 of G group DPB101:01:01G has the TCE group assignment 1 (before it was unassigned). Thus, the TCE group assignment for G group DPB1*04:01:01G and DPB1\01:01:01 would no longer be unique (group 1 and 3), which led to the ambiguous score (‘X’) in Hap-E search results.

We consider this to be an error in the data basis of HLA release 3.53.0 and have corrected it so that both G Groups again has the unique TCE group assignment 3.

Thus, Hap-E now behaves again like the IPD-IMGT/HLA.

P.S: Screenshots to IMGTHLA data:

[zabeen]

Solution

The issue has been resolved on both UAT and LIVE-WMDA-ATLAS as follows.

Fix

I manually edited the assigned TCE group for the following DPB1* typings in v3.53.0 of the Atlas HLA metadata dictionary, from 1 to 3:

• 1148:01, 1148:01Q, 1148:XX

• 1038:01, 1038:01Q, 1038:XX

I then manually refreshed the matching algorithm cache to propagate the changes.

Testing

Using the patients listed above, I ran “before” and “after” searches, and observed the following results which are in line with the expectations listed in the ticket description:

Patient 1) 01:01:01, 04:01:01 Donor 35512717 - 04:01:01G, 16:01:01G - Unknown -> PermissiveMismatch Donor 26911245 - 01:01:01G, 16:01:01 - Unknown -> PermissiveMismatch Donor 31594122 - 01:01:01G, 02:01:02 - Unknown -> PermissiveMismatch

Patient 2) 04:01, 04:01 Donor 957424 - 1:CMVHN, 04:CMVHV- Unknown -> PermissiveMismatch Donor 7900072 - 01:01:01G, 10:01:01 - Unknown -> Mismatch, NonPermissiveHvG

Patient 3) 01:01, 04:01 Donor 350221 - 04:01:01G, 13:01:01G - Unknown -> PermissiveMismatch Donor 984775 - 04:01:01G, 06:01:01G - Unknown -> Mismatch, NonPermissiveHvG

Risks

• On next release of the IMGT/HLA database (scheduled to be in approx. 2 months), if the above alleles still have TCE group of 1 assigned, then the manual data fix will need to be reapplied after the data refresh job completes.

[RimkePostel]

Hi @zabeen, I checked my searches: the issue has been resolved! Thanks a lot :)