How does nanopore allow targeted sequencing of large genomic regions?

[JackieBadger]

An advantage boasted by oxford nanopore is that no PCR is required. However, it is not clear to me how one would target a large genomic region, say 1MB for arguments sake, without a huge PCR at a low number of cycles?

Thinking "nanopore" is new to me so please excuse me if the answer to my questions is elementary.

Cheers,

J

[singlemoleculer]

One advantage to nanodetector technologies like Oxford and Nabsys is that they are single molecule so you look at individual molecules rather than large pools of identical molecules. Typically, you would not target a specific region but would look at the entire sample that would include the whole genome. I am not aware of any methods for efficiently targeting very long DNAs (other than flow sorting chromosomes or the like). If throughput is high enough, that is not a problem. If throughput is low, it is a problem.

[JackieBadger]

This is what I thought.
So it essentially relies on being able to sequence an entire genome, or chromosome, that contains your region of interest.

What is the maximum sequence achieved thus far using nanopore seq?
Last I heard it was ~48KB

[singlemoleculer]

I think you are referring to the Oxford data that was discussed but not really shown at AGBT12. At AGBT14, more detail was given though still limited. Average read length in the data described by Jaffe was 5 kb with the long tail up to about 10 kb. Nabsys uses a mapping rather than sequencing approach so it is used in a complementary fashion with short reads. Base-by-base data from short reads is combined with long range information from single-molecule nanodetector traces. Nabsys showed molecules up to 150kb in length. Posters of this are online: http://www.nabsys.com/News-Events/Pu...entations.aspx

[JackieBadger]

thanks!

j

[seqsense]

Hey, don't forget that unless the sequencing is done on a single cell, which does happen for researching tumour genesis, there will be many, many copies of the DNA already without the need for PCR given the multiple cells that will be extracted in sample preparation.

Besides, there's been comment about the nature of the T issue being a systematic problem that bioinformaticians can't handle well, but PCR always introduces systematic errors by its very nature and also drowns out any hope of identifying any other bases except A, C, G and T.

But now there are five different C's! What else don't we know.

[JackieBadger]

Besides, there's been comment about the nature of the T issue being a systematic problem that bioinformaticians can't handle well, but PCR always introduces systematic errors by its very nature and also drowns out any hope of identifying any other bases except A, C, G and T.

But now there are five different C's!

I'm sorry but I don't follow any of this.

What is "the T issue"?
What evidence is there to support that there are more than five nitrogenous bases and how does PCR suppress this evidence?
What are the "five different C's"

[singlemoleculer]

I can't comment on the T issue. No clue either. As for more than 5 bases, there are hordes of different modified bases that vary with species. In humans 5meC is most common but that is not true for all species. The next most common in humans is probably 5hydroxymethylC. PCR obscures this because the history of modification is lost when the DNA is copied multiple times during PCR. Single-molecule systems like PacBio or Oxford can, in theory and sometimes in practice, detect some of these because the modified bases are detected or copied directly. As for multiple cells being used for sequencing, that does not eliminate the need for PCR with non-single-molecule systems because each molecule has to be identical but each cell's DNA will break in unique ways. PCR is needed to make the ends the same. And yes, this can and does introduce issues with amplification bias and sometimes copying errors.

[JackieBadger]

OK. It wasn't immediately clear that he was refrying to modification of bases by methylation. It will be interesting to see if we can harness this information in DNA sequencing down the road, and the implications that it may have.

[seqsense]

So far, apart from C, MeC and HOMeC, there's also evidence of 5-carboxy-C and 5-formyl-C and their roles in epigenetics.

[krobison]

Oxford is probably a poor choice for sequencing targeted regions, given that most targeted sequencing is looking for small variations (such as SNPs) and Oxford looks out-of-the-box to not be well suited for this. Also, all of the existing capture schemes are designed around small fragments suitable for short read sequencers. It would seem that capture sequencing is playing to none of Oxford's strengths and against many of its likely weaknesses.