nanoporetech / dorado

Oxford Nanopore's Basecaller
https://nanoporetech.com/
Other
491 stars 59 forks source link

How much better is `dorado correct` in 0.7.2 than 0.7.0? #896

Closed sivico26 closed 3 months ago

sivico26 commented 3 months ago

Hello there,

I ran dorado correct 0.7.0 (actually 0.7.1rc1) on a big dataset (75x of ~ 4 Gb plant genome). Dorado 0.7.2 includes 3b51c1b, which I imagine improves read-error correction. The question is: do you think it is worth to rerun the error correction? is the difference significant? Do you have better qualities overall? Does this impact assembly in your hands?

For my dataset, version 0.7.0 took 191 hours, so I prefer to ask before launching and waiting again.

iiSeymour commented 3 months ago

Hi @sivico26

No, the difference is not significant so I would not recommend rerunning with v0.7.2 vs 0.7.1rc1 as these were only minor changes and we are still actively improving correction. We will make it clear when we believe there is a significant benefit in future releases.

sivico26 commented 3 months ago

Thank you for the quick reply @iiSeymour!