Normalisation (CLR, TSS or NONE) for coverM abundance

Downstream analysis and statistics MaAsLin
1

Hello MaAsLin3 developers and users,

I have metagenomic dataset and the abundance is trimmed mean from coverM that has been normalised by sequencing depth. I have filtered to remove samples with rowSum=0 and taxa with colSum=0.

Am I correct to assume that I use normalisation= NONE or should I still use normalisation for this type of data?

Thank you!

2

Hi,

Is there a reason you’re using trimmed mean rather than relative abundance from coverM? If your input data is already relative abundance (or something equivalent), you can use normalization="NONE" and still get results interpretable in terms of relative abundance.

Will

3

Hi Will,
Using trimmed means is standard we use in the lab as we want to get the per genome(MAG) coverage. I can perhaps calculate relative abundance from the trimmed mean values and use those as input?

Thanks you!!

4

Assuming you’re using this, I’d use the relative abundances since it looks like the trimmed means doesn’t account for genome size.

Will

5

Yes that’s it.
Alrighti, i’ll use relative abundance then!

Thanks Will