We are trying to decide between Illumina and 454 technologies for a MHC genotyping project in a bird species (class II B exon2). After running a preliminary run on Roche Junior and carefully filtering out artefacts, we found we need a very large sequencing depth to saturate the variability of alleles in each individual (up to tens of alleles). For the full experiment of genotyping 200+ individuals I would prefer MiSeq over Titanium given the price tag. But amplicon sequencing may get tricky there (low complexity problem) and I heard about huge headaches when analysing such MHC data from Illumina. The only related discussion I found on RG is this:
https://www.researchgate.net/post/Are_there_any_strategies_for_sequencing_a_polymorphic_locus_eg_MHC_without_using_cloning
But it does not quite address the problem.
Is anyone doing this? How are you making your amplicons? One big problem with sequencing low plex amplicons is noise. With the SBS chemistry, you get a signal from each cluster that incorporates the most recent base adding to the chain. So, when there are several similar clusters, then everything is lightting up at the same time, and will mask other sequences. The solution to this is adding PhiX during cluster generation, but you lose some seqeuncing space. The other issue with MHC is that it is highly repeditive, so designing primers or probes for amp/capture becomes difficult...which you probably already know. I'm eager to see what others have to say about this.
Thanks for your comment, Casey. Low complexity is certainly a problem and a lot of PhiX would be needed. Other approach is designing primers of different lengths to diversify the sequences on clusters. Both were tried on MHC amplicons by one colleague of mine. I was told it helped to increase the quality of reads, but still not to a level that would allow unambiguous allele calling. I was hoping to hear more opinions on this either here or on Seqanswers.com, where I also posted the question. But it seems not many people experimented with MHC and illumina so far.
For now, I will probably go with IonTorrent, which has similar pros and cons as 454 but with much higher no. of reads.
The repetitiveness of MHC is what makes allele calling so difficult with Illumina. It seems that you may get some true alleles (or even pseudogenes) with very low frequency, which is close to the error rate of your reads.
Jan,
Analysis of macaque MHC alleles is currently performed by some lab using Roche 454 and Illumina technologyies. In these species, MHC is complicated by multiple duplication of MHC-A and MHC-B loci, with sometime more than 19 functional genes per haplotype. The group of David O'connor has published several interesting papers on the subject and a recent paper may be of interest to you.
Dudley DM, Karl JA, Creager HM, Bohn PS, Wiseman RW, O'Connor DH.
Full-length novel MHC class I allele discovery by next-generation sequencing: two platforms are better than one.Immunogenetics. 2013 Nov 16. [Epub ahead of print]
Bernard, thanks for an interesting paper. I'll go through it. Mapping shorter paired-end reads with higher seq. depth onto longer 454 seqs looks neat. In my case I'm not even trying to reconstruct haplotypes, just looking into overall allelic diversity, positions under selection, etc. Because I'll be genotyping birds, where the situation is usually even worse than in mammals - high level of multiplication, I didn't think of trying to reconcile alleles into haplotypes. But it sounds very tempting for future.
Just an update. In the end I decided to go for IonTorrent - more bang for the buck than with 454, and one can use similar pipelines to process the data. But recently I was told a new paper will be out soon in Mol. Ecol. Resources on the MiSeq genotyping protocol of MHC. So those interested, keep an eye out for it.
Hello Jan..Currently we are going through the same decision, we want to do some MHC for bats and we are deciding which platform to use given the cost/quality of data. It seems you are using the IonTorrent, Is it a big differences in prices between this one and the 454 Roche? Are you using any company to the lab preparation?
Hi Yessica, we were lucky to use an in house facility at another local university. Hence, we are paying only the chemistry, which actually makes the runs much cheaper than regular 454. But there are downsides too. With my undergrad student we've just received the data and didn't have time to look into it thoroughly, because of preparations for a soon to leave to field trip. We've got several times the volume of sequences that one would get from a Titanium for fraction of the cost, but the quality of the data seems lower than from 454. But as I said we'll only look into it in detail yet after returning from the field trip.
Hi Jan,
How have things gone in the last half year?
We are using the FLX to type human HLA. Since the instrument will no longer be supported from mid 2016 on, we are working on switching our assay over to MiSeq. A number of labs are already successfully doing this.
Here is an example:http://www.biomedcentral.com/1471-2164/15/63
"Cost-efficient high-throughput HLA typing by MiSeq amplicon sequencing"
Hi Franziska,
we decided for IonTorrent and already processed the data, but we're switching to MiSeq with our next MHC project too. The Iontorrent was definitely workable and much more cost effective than 454, but suffered relatively high error rates. Thanks for the reference, I've seen a few Illumina MHC papers recently, not this one though. Best, Jan.
- Badges
- Science method