I want to classify metastasis and non metastasis samples. I am using "gene expression RNA-seq - STAR - Counts" data provided by https://xenabrowser.net/.It data has following features:
type of data gene expression RNA-seq
unit log2(count+1)
platform Illumina
wrangling: Data from the same sample but from different vials/portions/analytes/aliquotes is averaged; all data is then log2(x+1) transformed.
input data: formatROWs (identifiers) x COLUMNs (samples) (i.e. genomicMatrix)
Trimmed Mean of M-values (TMM) normalisation is something that is applied to log-fold-changes, not to expression values from individual samples. Effectly you calculate the change between two conditions and then normalise that change. Since the analysis you are applying here doesn't involve two conditions, then TMM isn't applicable.
You'd find this very quickly if you actaully tried to apply it!
Personally, I use rLog or VST for machine learning applications (although you couldn't do that on log2(count+1) data, you'd have to convert it back to count), but I think there is also a way to use voom.