I have metatranscriptomics data set accompanied with metagenomic data (metagenomic and metatranscriptmic data obtained from the same sample).
I know how to use humann3 for functional profiling of shotgun metagenomic data, however, I am not sure how to use humann3’s in-built options for metatranscriptomics analysis or a combined analysis (metagenomics + metatranscriptomic).
Is there a tutorial available on how to use humman3 for this purpose?
To actually analyze the metatranscriptomic (MTX) reads we recommend using the taxonomic profile from the sample’s corresponding metagenomic (MGX) reads. You can pass this to HUMAnN via the --taxonomic-profile flag. This will skip reprofiling the species from the MTX data.
For downstream analysis, we have some recommended code for that here:
This is based off of our findings from this paper:
MetaPhlAn 4 will work for MTX reads, but we don’t generally recommend taxonomic profiling from MTX as the resulting abundances are difficult to interpret. They mean something along the lines of “What was the average relative RNA level of this clade’s unique sequences relative to other clades’ unique sequences?”
As additional challenges: 1) these values will be confounded with the clade’s DNA abundance; 2) there is no guarantee that the values are directly comparable between species (as, say, a measure of activity); and 3) variability in marker expression could also differ between species (whereas at the DNA level we expect marker coverage to be constant within species except as influenced by read sampling noise).
I am a novice in metagenomic / metatranscriptomics analysis and slowly progressing on the steep learning curve.
I came across your article “Statistical approaches for differential expression analysis in metatranscriptomics” and since I have both MGX and MTX data from the same samples, the methods you propose for normalizing the MTX data with paired MGX data is something I’d like to apply.
As my data is from soil samples, the taxonomic resolution is not great. Therefore, I’d like to apply either the RNA/DNA ratio model or the Feature-DNA covariate model to transform the MTX abundance data. I’ve looked at the MTXmodel library R package and ran the Demo. However, I don’t really understand what the MTX model does nor the output obtained.
What I’m trying to do is transform/normalize the MTX abundance data with one of your models to then be able to run differential expression analyses on this data.
Could you guide me through this process ?