Variant calling on synthesized DNA

Question

[stan.aanhane]

Hi everyone,

I want to ask your opinion on the following subject:

Currently, we are producing synthesized DNA with the use of Multiply-Primed Rolling Circle Amplification. This allows for the production of a lot of DNA. Within that DNA there is a region which can be used for gene therapy. To be sure that our DNA polymerase didn't introduce too many errors (the error rate is 1 for every 1.000.000 bases), a QC has to be made. We want to use MinION for this. Now we are thinking of 2D sequencing (using our own caps to achieve this). Next to that is the data analysis, which is a bit more complicated due to the fact that most SNP variant callers are used in WG analysis. What I would like to ask is, what do you guys think of the 2D sequencing step to improve accuracy? And what are your suggestions on the data analysis? We are currently thinking of P.E.P.P.E.R. (https://github.com/kishwarshafin/pepper). But this seems like a bit overkill maybe.

I am looking forward to what you guys think.

Thanks!

Answer 1

[colindaven]

Definitely use the new Kit14 nanopore 10.4.1. The error rate is a lot lower than on the older 9.4.1 chemistry. This will help SNP calling.

I would look up duplex sequencing from Nanopore and do this before you try to implement your own 2d sequencing. AFAIK it only works for 5-70% of reads depending on if you believe the company or not, but it will be far quicker and more cost effective than chemistry development.

You can start SNP calling analysis with the tools longshot (SNPs only, no indels) and Clair3. I think Pepper is overkill and not very performant or intended for real usage (at least genome wide), but maybe it would work on your dataset.