multisplit and or single-sample

[sheaster]

thanks for a great tool!

would there be any benefit to running VAMB with single-sample assembly and binning as well as running VAMB in multisplit mode with differential coverage information included (submitting both sets of bins to dastool for that metagenome)?

or would one expect that multisplit results would be superior and sufficient?

Thanks,
Shea

[simonrasmu]

Hi Shea,

Multi-sample (multi-split) is preferred for any binning as the co-abundance vector (abundance of contigs across samples) is a very powerful determinant for binning. So with VAMB always use this.

However, with our newest binner TaxVAMB single sample binning is a lot better than it used to be and can rival multi-sample binning - have a look at the pre-print: https://www.biorxiv.org/content/10.1101/2024.10.25.620172v1

Best,
Simon

[sheaster]

Hi Simon,
Thanks for the response. I will take a look at the pre-print!

This may be a naive question, but how do you conceptualize when it is appropriate to multi-sample bin?
I see in the original VAMB publication, many human gut microbiome samples are run with multi-split - all from a similar environment.
But, for example, how about from different depth in a vertical profile of a water and sediment column (from a lake for example)? Or from different soil sampling sites?

Thanks,
Shea