2
$\begingroup$

I have a couple of metagenomic bins which are okay in quality, but often missing rRNAs (16S, 23S...). I assume this is due to high population variability, combined with the high conservation of rRNAs, and reads of at most 300 bp. My binning software of choice is vamb.

Could this be fixed? Once a metagenomic bin is built, is there a way to see which 16S (or multiple 16S possible sequences) are likely to be associated with it? I considered using tools which explore the assembly graph, either manually (Bandage) or automatically (spacegraphcats), but I'm not sure if this is the right approach. After all, if the rRNA genes were linked in the assembly graph to genes that now clearly belong to a bin, they would have been added to that bin...It feels like this is a problem others may have encountered before, so I was wondering if anyone has an idea : )

$\endgroup$
3
  • $\begingroup$ I am not super familiar with vamb, but it doesn't appear to use the assembly graph: "Vamb is a metagenomic binner which feeds sequence composition information from a contig catalogue and co-abundance information from BAM files into a variational autoencoder and clusters the latent representation." Most binners I'm familiar with don't pay attention to the assembly graph. I'd expect that there is some assembly graph signal, possibly the issue is that there is a really nasty knot making it hard to navigate the graph properly. $\endgroup$ Jun 16, 2022 at 17:13
  • $\begingroup$ @MaximilianPress Sorry, I was confusing. I meant using the assembly graph from my assembly program (MEGAHIT) and then trying to correlate the contigs with the bins. If you'd like to expand on the knot it would still be helpful! ('here are the reasons this would be difficult' is a valid answer : ) $\endgroup$
    – Laura
    Jun 16, 2022 at 17:33
  • $\begingroup$ Ah, I mostly meant that "After all, if the rRNA genes were linked in the assembly graph to genes that now clearly belong to a bin, they would have been added to that bin..." isn't necessarily true if your binner doesn't pay attention to the graph. In conclusion: yes please do use spacegraphcats or a similar tool. $\endgroup$ Jun 16, 2022 at 18:07

1 Answer 1

2
$\begingroup$

I considered mentioning this but considered it implausible. However, apparently people are actually able to use PE shotgun data to link rRNA genes to MAGs.

This is something that is quite trivial with linked-read or Hi-C data, but I am assuming that you don't have these data (and don't have the desire to go back and generate them). However you presumably have PE shotgun data that you used to generate your MAGs.

$\endgroup$
1
  • 2
    $\begingroup$ Really impressed that you linked something so recent, thank you! $\endgroup$
    – Laura
    Jun 23, 2022 at 12:37

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.