When you say polymorphic, do you mean repeat regions? If, for example, there are 6 400bp repeats in a row in one individual and 8 in the next individual you will have a very difficult time using current NGS library technology (which typically chops DNA down to 200-500bp per clone) to determine the exact number of these repeats. We are dealing with a similar problem now and are looking into two newer technologies:

1) PacBio sequencing, which uses much larger clones in the sequencing library so that you will most certainly sequence through non-repeating sequence in the same clone allowing you to align that sequence in the genome (http://www.pacificbiosciences.com).

2) Oxford sequencing which uses nanopore technology to sequence extremely long linear segments of DNA (http://www.nanoporetech.com).

The problems with these two new methods are that they yield MUCH less sequence than current NGS. To my knowledge there are currently no methods to enrich for a genomic region using these two platforms, but I hope they fix that soon.

This may not solve your specific problem, but I think that people should be aware of these new methods since they put a different spin on whole genome sequencing methodologies.

Good luck with your experiments.

Next Generation Sequencing can be applied to PCR products including the targeted polymorphic region. After PCR, the samples can be barcoded and pooled for sequencing. NGS systems have their specific library preparation protocols and barcodes for this step. Also, primers including different barcodes and adapter sequences of NGS platform can be designed and amplicon sequencing can be performed with the PCR products of these primers. Hundreds of samples can be sequenced in a single run but the depth of coverage and pooling steps have to be evaluated carefully. Different NGS platforms can be used according to the amplicon size of the targeted region. Since each platform has its own advantages and disadvantages, the choice of the platform has to done carefully.

As far as I know, FLX 454 has a custom protocol specifically for HLA typing (http://454.com/products/assays/hla.asp ) but I have no idea about the efficiency of the protocol and data analysis. If you have a lot of samples, using NGS approach can be handy and cost effective.

NGS has been helpful in determining percentages of mutant alleles vs. wild-type alleles in our experiments. For example, in a patient, we had discovered a germ-line mosaicism in the person's TP53 gene with the aid of NGS. After NGS analysis of the person's lymphocyte DNA, the number of mutant reads was 20% of the total sequences (PMID: 22006311). To confirm this 20%, we subcloned the DNA and Sanger sequenced the DNA clones. After Sanger sequencing, we found that the percentage of mutant allele was 21.5%, which was similar to 20% obtained from NGS.

When you say polymorphic, do you mean repeat regions? If, for example, there are 6 400bp repeats in a row in one individual and 8 in the next individual you will have a very difficult time using current NGS library technology (which typically chops DNA down to 200-500bp per clone) to determine the exact number of these repeats. We are dealing with a similar problem now and are looking into two newer technologies:

1) PacBio sequencing, which uses much larger clones in the sequencing library so that you will most certainly sequence through non-repeating sequence in the same clone allowing you to align that sequence in the genome (http://www.pacificbiosciences.com).

2) Oxford sequencing which uses nanopore technology to sequence extremely long linear segments of DNA (http://www.nanoporetech.com).

The problems with these two new methods are that they yield MUCH less sequence than current NGS. To my knowledge there are currently no methods to enrich for a genomic region using these two platforms, but I hope they fix that soon.

This may not solve your specific problem, but I think that people should be aware of these new methods since they put a different spin on whole genome sequencing methodologies.

Good luck with your experiments.

I have to chime in support for the use of NGS. To look at MHC sites probably the best platform will be 454, because it will most reliably support longer sequence reads, i.e . the read assembly would be unequivocal. The emulsion PCR process, required for pyrosequencing does a fine job of separating the clones. Each patient can be tagged individually either by post PCR ligation of the barcoded adapters, required for sequencing, or better yet – by using MHC universal primers that have been made to include in the 5’-end the needed adapter and barcode.

Cost example – 454JR system – for one thousand dollars one can get up to 70000-100000 individual reads (average read length of approx. 400bp, BTW the built it control fragments work reliably at 500bp). One can easily multiplex, e.g. 10 individuals in one run with ten different bar codes and thus get 7-10k sequences per sample. The GS-Flex will give 10-20 times the amount of data for about the same cost. The Ion Torrent platform is a good, a bit less expensive, alternative and offers up to 6.5 million reads, however at this time the read length is maximum 400nt.

If NGS is to be omitted entirely, end-point PCR is very good, sensitive and accurate method for isolation of different molecules without biological cloning. The positive clone selection and sequencing are easy to automate to achieve a good throughput. However, the required Sanger sequencing after the clonal separation is making the process extremely costly (total cost- about $10 per clone).

There are many types of NGS approaches that can be use to tackle your problem. As Muzzafer stated, 454 has a protocol for this type of sequencing on their website. Illumina also has some "out of the box" kits available that may be useful for this that are extremely user friendly and require little experience: http://www.illumina.com/truseq.ilmn. I think the most difficult part of these types of projects is designing the initial barcoding primers for your gene of interest (MHC). MHC is also quite a large gene, so if your amplicon is larger than 400bp, 454 will be a better platform for sequencing.

Illumina platforms, particularly if you find a facility with a Miseq, tend to be much cheaper than 454 systems, but it will depend on your facility. The cost will depend on whether you're developing all the methods yourself (i.e. designing your own barcoded primers, amplifying, pooling, gel purifying) or if you purchase some type of kit (Illumina and Agilent make a variety of kits for these types of projects, but make sure they are compatible with your sequencing facility). It could range from $10-$200 per sample... but the number of reads you need also must be factored into the $/sample price tag.

Most sequencing facilities will provide a wealth of information on their websites about how to prepare libraries for sequencing and tend to be pretty helpful in assisting you with designing your pool. Also, depending on how many samples you plan to pool, whole-genome sequencing has become quite affordable for up to 25 clones and could also provide the information necessary to evaluate polymorphisms within your gene of interest. However if you have 50-200 clones, multiplexed PCR amplicons would be the best route.

Dear Brain,

I dont want to make any comment on sequencing a polymorphic locus by Next Gen Sequencing approaches. But, It will be good if you clone your PCR product (which is having a polymorphic sequence) and screen your colnes by PCR-RFLP. That will save your time and money and tell you about polymorphism among the clones. Now you select few clones (depending upon the polymorphism) for sequencing.

HLA/MHC region is the most tricky region not only because it is highly polymorphic but also because of the adjacent pseudo genes. I tried designing primers for the 2nd and the 3rd exons of the Class II loci DQB and DRB for my doctoral work and it was so difficult to get specific amplification. Primers that seemed perfect when testsed with various softwares gave multiple products of different sizes(when the PCR product was run on a 2% agarose gel). I always wondered if the pseudogenes also got amplified. I took primers from a published paper and sequenced the samples but some of the samples could not be sequenced.

I doubt if NGS will be able to overcome the problem of pseudogenes. Polymorphism in this region is also population specific so unless NGS library is designed to overcome these problems it will not be of much help. 454 seems to have protocols to address these issues.

Thanks everyone for the very useful answers so far.

To help simplify the question further, my work was on the salmonid MH class I and class II genes. These fish have just single, very polymorphic class I and class II loci on separate linkage groups. So, I would have a single PCR product with at most two polymorphic products. Normally, I would clone these, take five clones and sequence each. With automation (my last post-doc was setting up this http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0048022), as is, I think we could do 1,440 DNA-extracted clones (288 fish) in about 2.5 days (cleaned sequencing products are run externally). Around €13 per individual in total.

Speaking to a couple of colleagues there who have used NGS approaches across multiple individuals/loci, the number of primers involved can be costly and it might not be suitable for screening across large numbers of individuals.

Might automation and older fashioned molecular biology be better for certain population genetics projects?

Incidentally, about €6 of that €13 cost is cloning so what I was really hoping for was some strategy that avoids the need for cloning?

Brian, may I suggest you try something in between? It looks like you could use a custom SNP array method (Agilent (http://tinyurl.com/AgilenArray) will make you an array). Using this array across the MH class I/II region you will immediately get information about all of the polymorphisms that are known (unknown would be harder to do, but could be done). If you want to screen a LOT of DNA samples for a small region, the array method is the way to go and does not involve cloning.

Thank you for the contributions thus far.

I know people used DGGE far back to separate out different sequences but it's obviously messy.

my two cents would be to use PCR based detection

on similar lines to

http://www.biotechniques.com/BiotechniquesJournal/2006/November/Real-time-PCR-typing-of-single-nucleotide-polymorphism-in-DNA-containing-inverted-repeats/biotechniques-40712.html

Now, I would like to ask Brian's question but with a twist: how about in a non-model species with only a confamial species whole genome sequenced? I have not a priori knowledge of how many duplications or alleles to expect. Also can the difference in functioning genes and pseudogenes not be determined by sequencing both cDNA and gDNA?

Any paper where MHC is sequenced without cloning or NGS?

In principle, NGS is suited to do polymorphism analysis - Illumina or 454, both were applied for this purpose in our group and by collegues. However, it is really large-scale and requires that you have ~1000 samples, or large regions in a few individuals (combined with some target enrichment method). It is less accurate than Sanger sequencing which may drag on calling reliability. If you do multiplex PCR, then carefully think about the possible formation of chimera PCR products. I recommend to use Sanger for primary polymorphism characterization whenever possible. Based on such primary data, NGS is absolutely fine for genotyping.

An example, Illumina NGS for sequencing a CDR3 (antibody) library:

NAR 38:e193, doi:10.1093/nar/gkq789

With regard to the PCR primer design challenge, as Aruna pointed out. Microarray-based target enrichment is a way to get to your locus amplification-free. This also offers a way to work on novel species, given some homologous sequence.

I am planning a similar project and was very interested to read the above about Haloplex. However, from what I can gather, a 96 sample custom Haloplex kit for the 1-500 kb product range costs ZAR 375000 (I am in South Africa). That is USD 38000!!

I definitely don't have that sort of funding. Just a heads up-- this method does not seem to come very cheap.

A PCR/Sanger aproach can be used. In this case you sequence first both alleles together, you get a mixed sequence from both alleles. Knowing the mixed bases you can develope sequencing primers on either of the two mixed bases, and you can sequence one of the two alleles in this way, your sequencing primer should not be longer then 10-11 bp, to get specific sequence from one allele. You can use the same PCR product for this, you only have to sequence it twice.

See Stutz and Bolnick PLoS One for an example