Correcting ONT reads - Alternative to vechat?

Question

[Crying_monkey]

Hello,

I have been trying to use vechat but without success to correct my ONT reads, and the github seems unresponsive the last release was in 2022. I was curious to know what tools you use to correct ONT reads with only ONT reads, no illumina or PacBio are available here. I would like to know what tools the community uses. Or even, if you don't correct the reads, your voice is also interesting.

Thanks a lot

Answer 1

[GenoMax]

If you have access to the entire run folder you should use ONT's basecaller dorado (LINK) to generate super accuracy reads. After that dorado has an option dorado correct that you can use to polish the calls.

Answer 2

[colindaven]

What is your goal and use case ? SNP calling? De novo assembly ? etc

Correction with ONT using dorado correct is only useful for de novo and uses a lot of resources. Maybe 64 cpus, >800GB RAM and then a gpu for 2-3 days depending on your dataset size and coverage.

If you do not have these resources, you might be better just using Hifiasm with the new --ont feature flag.

Other ways of correcting ONT reads are Canu (also resource hungry). Also hifiasm corrects reads itself - but I don't think it outputs corrected FASTQ reads (maybe thats an option somewhere). Lastly, there is dechat - https://github.com/LuoGroup2023/DeChat