Human DNA methylation running on MinION

Question

[ResearchR]

Dear all,

since I know that some very knowledgeable people concerning Oxford Nanopore Sequencing Technology are members of this forum, I will just leave my lack of experience here (and try to learn something). Following the situation: I am supporting a humanitarian project in Cambodia, which primarily deals with diagnostics based on histopathology. We had some success implementing machine learning based tools improving and accelerating diagnostic decision making. Now I was asked if we could add genetic information to the histopathological images, since this (of course) is of advantage when it comes to cancer, but also other diseases. We would be especially interested in DNA-Methylation. Surprisingly, I was rather rapidly confronted with Oxford Nanopore MinION and VolTRAX (well...I was asked what the "self-sequencing" technology is called) and since it seems cost effective, which would fit with the grant application. Since I only worked on Illumina technologies, here my thoughts/ struggles:

1. PromethION is financially not feasible, so we are left with MinION or GridION. According to the spec-sheet, a flow-cell can produce up to 50Gbs, which would be something like a 15X coverage. If I think of WGBS in short-read terms that might not be sufficient or really borderline.
2. A basic laboratory is available, which would allow PCR amplification and DNA extraction. Does the VolTRAX really perform as promised in the advertisements and would be a great addition or is the technologies in its infancy?

Oxford Nanopore would really be a nice option, since the sequencing process does not rely on bisulfite treatment (or did I miss something?) and we would have, in addition, the raw read information. The only alternative I could see would be Illumina 850k arrays.

So, if someone could comment on my lack of knowledge/ experience, I would really appreciate the effort!

Cheers and have a sunny day!

Answer 1

[colindaven]

I have performed analysis on human minion sequenced data. It was a long time ago, so was only 3.5X I believe. Link: https://www.mdpi.com/1422-0067/22/8/3937

For bacteria, I have seen 20 GB per flowcell quite frequently. For human and for poor clinical samples, expect a lot less. I think 10 GB would be "good". But that is only 3.5 X for your experiment, so similar to ours.

Nanopore is a lot more robust now, but if you don't have a lot of experience your yields will vary greatly. Can be due to the flow cell quality too.

ONT does not rely on bisulfite.

Illumina arrays or EPIC sequencing might be an alternative if you need more guaranteed values across your cohort, but nanopore is a lot more fun :-)