Errors in certain HLA alleles

[bgrumbt]

Hi,

we adapted HLA typing (amplicons of exons 2+3 with own primers) from Roche 454 to MiSeq. Now I have the problem that certain DRB1 alleles show many errors in the forward read 1 for exon 2, although I have perfect coverage of reverse read 2. And other similar alleles amplified with the same primers do not show those errors and have good coverage.

Has anyone seen similar problems, that a certain sequence somehow can't be sequenced?
Could it be that PhiX interfers?

I don't now what else I can do?

Thanks for helping,
Barbara

[JackieBadger]

Hi,

we adapted HLA typing (amplicons of exons 2+3 with own primers) from Roche 454 to MiSeq. Now I have the problem that certain DRB1 alleles show many errors in the forward read 1 for exon 2, although I have perfect coverage of reverse read 2. And other similar alleles amplified with the same primers do not show those errors and have good coverage.

Has anyone seen similar problems, that a certain sequence somehow can't be sequenced?
Could it be that PhiX interfers?

I don't now what else I can do?

Thanks for helping,
Barbara

Do you mean you get lots of different variants per read 1 exon 2 which are just a few bases off the known true allele?

Base pair mismatch errors can be highly sequence specific, and amplicon specific. I see the same in MHC IIB data. So while some sequence specific errors are replicated among different individuals, some amplicons may have lots of unique repeatable errors (repeatable as in sequenced more than once).

if you want to recover all of your depth information you can correct the depth of the true allele through identifying repeatable putative errors. This is beneficial if you want to accurately delineate true alleles and contaminants, and to estimate copy number variation using relative sequencing depths.

We have a paper in review which deals with all of this. If you would like to read you can email me [email protected]

ps i have used "individual" and "amplicon" interchangeably

[bgrumbt]

Do you mean you get lots of different variants per read 1 exon 2 which are just a few bases off the known true allele?

Yes, I get only few reads 1 which match exactly to the expected allele and lots of reads with errors, many of which look randomly distributed, but some occur repeatedly in many reads (up to 50% of reads of one sample) and also in many samples. I blasted them, but so far I didn't find anything from which this mismatched bases could originate.

And what I don't understand is, that the other allele, which is amplified with the same target specific primers and which differs in only one base (!) is covered perfectly fine. And it is not this certain position, which is affected.

Thank you for answer,
Barbara

[snetmcom]

Can you give more details on the amplicon design and analysis? I assume these are overlapping PE reads?

[bgrumbt]

Hi,

yes it's paired end read, the overlap is 10 bp DRB1 exon 2 (human). We use dual indexing and the forward target specific primer is located 150 bp upstream of exon 2 in intron 1. We multiplex 192 samples in one MiSeq run and spike in 10% PhiX, because of low diversity (only 12 amplicons). The primer also amplify DRB3/4/5, but the other sequences don't come from these genes.

Regards,
Barbara

[JackieBadger]

192 samples in one run?
You should be getting huge depths!
This could explain why you see so many different errors.
What do your depths look like?

In a similar situation, where for instance I sequence an amplicon at 30,000x I see lots of repeatable errors. Still, the "true" alleles can easily de distinguished.