Duplex Sequencing on MHC

[jpleonard2000]

I am interested in sequencing a small fragment of the MHC and am researching Next Generation techniques that promise to reduce error. The standard protocol for sequencing these genes involved a tremendous amount of cloning and Sanger sequencing.

Does anyone have any experience conducting Duplex sequencing as described by the article below?

http://www.pnas.org/content/109/36/14508.full

This technique was developed to look for very rare mutations, but it could also be used to find MHC alleles which are known to have multiple repeats in an individual. Any thoughts?

[missle]

hi~ I'm also doing this work now. But so far, I didn't get a positive result. There always were some adapters ligated with each other while I wanted to ligate the adapter & target DNA. Have you finished your work? How was your results? Do you have any suggestion for me? Thx~

[JackieBadger]

Couple of things...

first. Base mismatch errors associated with NGS can be sequence and base position specific, and thus repeatable. Sequencing both strands of the DNA would not overcome this error.

Second. because the method relies on double stranded sequence to "verify" mutations, I would be very very cautious of it. i.e it would accept repeatable base mismatch errors seen at low depth (because as far as I can see from the abstract, the method does not use depth to distinguish real alleles vs artifacts)

third. Low copy number cross sample contamination DOES HAPPEN FREQUENTLY during DNA extraction (especially if you are using a robot that flushes tips), and during PCR. Unless you already know the allele number of multitemplate MHC fragments, you are going to accept potentially high numbers of contaminants (unless you are uber careful during sample prep).

forth. Why do you need to use a method to detect "rare alleles"? In MHC you will expect 10s of alleles max, all in high copy number in your organisms, and so would be sequenced many times

Fifth. I struggle with these methods to detect "rare" alleles. If its real, it will be sequenced many times. I can see how an allele may be sequenced less when in a pool of many high copy number alleles, but otherwise I think these methods are a bit flimsy and probably full of error

[SNPsaurus]

Jackie, you make good points about things to be cautious about. But a rare allele will look like the prevailing error rate if it is less than 1% of the population.

Let's say you are looking for a "T" allele present at 1 in 1000 fragments in your population, compared to the predominant "A" allele. So you sequence to 10,000X coverage. Getting multiple reads won't help for these rare alleles, because every position in the genome will have 10 reads for T, but also for G and C because of the error rate. So you need ways to reduce that error to levels below the allele frequency.

[JackieBadger]

Agreed. I think I mention (although not in as much detail ) thats how I thought this approach may be useful. But for MHC genotyping, it is (usually) necessary to have individual genotypes. Thus you are sequencing per-individual amplicons. These are barcoded and pooled. No population level detection of rare alleles is needed.

[SNPsaurus]

I was just thinking through your fifth point a bit when I replied. It does seem like the rare allele methods aren't best suited for this application. Isn't getting an extended haplotype more important?

[JackieBadger]

Isn't getting an extended haplotype more important?

Exactly!

I think jpleonard2000 has heard/read about NGS errors, seen the difficulties associated with separating MHC haplotypes from errors in recent publications, and thought the rare allele method would circumvent this.
It wouldn't.

However, we are soon to publish a highly accurate and repeatable MHC NGS genotyping method that over comes such errors.

A difficult problem to address for sure!

[JamieHeather]

I can offer no insight, but I'd love to use a robust MHC NGS typing paper!

[JackieBadger]

Once the paper is accepted I'll post a link to it on here