I have a transcriptome that I would like to filter by expression value. Basically, I have 3 samples that I mapped against my transcriptome with kallisto.
I have then use Trinity tool abundance_estimates_to_matrix.pl) to produce an expression matrix, which gave me 2 matrix:
- a matrix of TPM expression values (not cross-sample normalized): 1st matrix
- a matrix of TMM-normalized expression values (or TMM-normalized TPM matrix): 2nd matrix
My first idea was to keep transcript (so contig) that have >=1 TPM in at least one sample (so using the 1st matrix).
Now I have doubt, should I use the TMM one instead ?? because results are a bit different.
Indeed, with the TPM values (1st matrix) I keep 30% of my initial transcriptome, while with the TMM values (2nd matrix) I keep 23% of my initial transcriptome
Thanks for your help.