Metagenomic Binning: before or after denovo assembly?

Question

[gatheringdusthere]

In dealing with NGS data from mixed samples (a sample that contains host and perhaps two or three bacteria of interest), would one perform metagenomic binning before or after de novo assembly?

I was taught to do binning before de novo but I am not sure if that is correct after some discussion with someone. It seems to make sense to do binning first, so reads are first categorized and then the reads from relevant 'bins' can be assembled. But it seems that reads that can't be categorized would be lost.

Would appreciate any clarification on this.

[Asaf]

binning based on what? Kraken assignment or something like that? What about the unmapped reads or reads mapped at higher taxonomic levels?

Answer 1

[Mensur Dlakic]

Both approaches are valid, although assigning the reads before the assembly is not something I would call binning.

I have read many posts on this forum advocating for removal of microbial contaminants before the assembly, and if you do some searching it will be easy to see why. I think one of the main reasons for that point of view is that people assume having contamination will affect the assembly, except that in most of those arguments microbes are the contamination. In your case it seems that the host DNA is what you want removed. Either way, genomes of eukaryotes (especially higher) and prokayrotes are different enough that usually there is no problem in co-assembling them. Doing the binning afterwards should be straightforward from contigs because the two groups are definitely different at the level of longer contigs. That is what I would suggest you do, although there are valid arguments for doing it the other way (or both ways and comparing).

A recent discussion on the subject:

Removing bacterial sequences from assembled eukaryotic draft genomes ?

[gatheringdusthere]

Thank you for your reply.