1
$\begingroup$

I'm working with whole genome metagenomic samples from human skin, and I'm using MEGAHIT for assembly and MetaQuast for evaluation. However, MetaQuast requires a list of reference genomes for the analysis (if I understood correctly). When I perform the evaluation without the references, it uses BLASTN for aligning contigs to SILVA 16S rRNA database. Since I'm working with whole genome sequences, this is not an optimal choice. As expected, MetaQuast is identifying only two species in my contigs file, although I have manually checked through BLASTN that there are contigs that match other species' genomes. And this evaluation makes my assembly looks terrible. Since I'm working with samples from human skin, it wouldn't be viable to download a database of reference genomes and perform MetaQuast using a local alignment. So, what would be the best approach to perform the evaluation correctly in my case?

Looking forward to discussing it!

$\endgroup$

1 Answer 1

1
$\begingroup$

I would suggest first using a binning approach such as MetaBAT2 (or an equivalent tool, there are many) to separate your metagenomic assembly into bins roughly corresponding to genomes. I would then run CheckM to assess the bins' quality/completeness/purity.

If you still want to use quast, you could then run metaquast with a reference genome appropriate to each bin independently. I would suggest GTDB-tk to perform a classification and identify such an appropriate reference genome.

$\endgroup$
2
  • 1
    $\begingroup$ Thanks @MaximilianPress good way to take the question forward $\endgroup$
    – M__
    Jul 26, 2022 at 21:19
  • 1
    $\begingroup$ Thank you @MaximilianPress, I received a similar answer at Reddit, so I believe this might be a better way to proceed indeed $\endgroup$
    – Poccia
    Jul 27, 2022 at 14:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.