search

GaetanBenoitDev / metaMDBG

MetaMDBG: a lightweight assembler for long and accurate metagenomics reads.
MIT License
79 stars 3 forks source link

Error threshold for usability with Nanopore reads? #8

Open Ge0rges opened 5 months ago

Ge0rges commented 5 months ago

Hi Gaetan,

Hope you've doing well since EBAME, I'm happy to see this tool gaining some attention.

I was wondering if you've formalized the maximum error rate tolerable by metaMDBG to function better than the existing methods (e.g. fly)? I'm specifically wondering when Nanopore reads might become good enough to make use of this.

Thanks.

GaetanBenoitDev commented 3 months ago

Hi Georges,

Sorry for the late reply. I hope you are doing well. I tested a bit nanopore data. Metamdbg can't process the raw reads at the moment (even with the R10 chemistry). But it can extract Mags if you polish the reads first (with self-correction or illumina data). I will try to handle the latest R10 data in a future release.

Ge0rges commented 3 months ago

Hello,

No worries. I'm curious what you mean by extract MAGs, I hadn't grasped that metamdbg included a binning algorithm? Or do you mean that it would only assemble very closely related contigs?

rrwick has anecdotally found that medaka polishing (I assume that's what you're referencing when you mention self-correction) doesn't have a significant affect on hac base called reads. Does that mean hac raw reads would work?

Thanks!

GaetanBenoitDev commented 3 months ago

With polishing, we could recover a few complete mag in a single contig, and a few mag after binning. You can find all details in the publication (we have added a nanopore section in the final version). But nothing from the raw reads.