Reader Comments

Post a new comment on this article

HLA A2 bias in class I restricted epitopes

Posted by ssunda25 on 04 Jun 2013 at 23:34 GMT

Hi, this is a very interesting article on bias in Class I binding epitopes. We were wondering what the distribution of HLA/MHC alleles were in the dataset you tested. Given the fact that there are a large number of epitopes defined for HLA A*02 than other alleles, have you looked to see if the bias is similar for other MHC alleles as well? Also, just to clarify, were the epitopes analyzed include epitopes from other organisms as well?

- Sri krishna, Diego Chowell, Biodesign Institute, ASU

No competing interests declared.

RE: HLA A2 bias in class I restricted epitopes

ykimbiology replied to ssunda25 on 06 Jun 2013 at 01:22 GMT

Dear Sri and Diego,

we very much appreciate your interest in our paper.

Here are responses to your questions:

1) The epitopes we used in the study are associated with a wide range of MHC restrictions. For vaccinia virus, the top 5 MHC alleles associated with its epitopes are as follows:

MHC NumberOfAssayRecords*
HLA-A*02:01 296
H-2-Kb 219
Unknown 132
HLA-A*11:01 101
H-2-Db 83
*An epitope may be associated with more than one T-cell assay records.

2) We agree that it would be interesting to extend the analyses to individual MHC alleles, but we could not do this because it would result in much smaller data sets to infer positional bias.

3) Yes, the set of viruses for which epitopes were analyzed included numerous organisms (total of 93 viruses). Of these, the top 20 viruses with the most number of epitopes are shown in Table 1.

- Yohan and Bjoern

No competing interests declared.

RE: RE: HLA A2 bias in class I restricted epitopes

ssunda25 replied to ykimbiology on 07 Jun 2013 at 01:23 GMT

Hello Yohan and Bjoren,
We really appreciate the rapid response, thank you. Perhaps we weren't clear enough; we were more curious to see if the over-representation of A02 epitopes in IEDB would bias the entire data set and that was the reason for HLA distribution question. In question 2, we were asking about epitopes from other "host organisms" and from your clarification it seems like you did.
- Krishna and Diego

No competing interests declared.

RE: RE: RE: HLA A2 bias in class I restricted epitopes

ykimbiology replied to ssunda25 on 11 Jun 2013 at 20:13 GMT

Sorry if our response was not clear: Yes, there could be a bias introduced by an over-representation of A02 data. We didn't see a strong biological rationale to expect an allele specific difference after not seeing significant difference in positional bias for binding motif distributions for different alleles, so we didn't subset the data further to test that. Given the data available, there are limits to how much subsetting makes sense, in particular as there are many other potential sources of bias in our dataset (e.g. over representation of proteins with certain functions as sources of epitopes). So if you have a nice hypothesis why HLA A02 should be different, we would definitely encourage you to see what you find.


- Yohan and Bjoern

No competing interests declared.