TMM.normalized.FPKM vs .TMM.EXPR.matrix for populating SQL

Question

[ando.kelli]

Hi All,

I'm using the Trinity/Trinotate pipelines (including RSEM/EdgeR), and ran the 'abundance_estimates_to_matrix.pl' step which generated two outputs as expected: 1) xxxxx.count.matrix and 2) xxxxx.TMM.EXPR.matrix

I'm further down the pipeline now, and want to upload my DE results to SQL. The script is asking for:

--fpkm_matrix Trinity_trans.counts.matrix.TMM_normalized.FPKM

None of the previous scripts have produced a .TMM.normalized.FPKM output. I thought of using the .TMM.EXPR.matrix file, but I read that the file (.TMM.EXPR.matrix) contains normalised TPM not FPKM at this site: https://github.com/trinityrnaseq/KrumlovTrinityWorkshopJan2016

Can anyone confirm whether the file .TMM.EXPR.matrix contains TPM or FPKM?

There is a script that comes with the Trinity suite run_TMM_normalization_write_FPKM_matrix.pl) that may generate the file I need, but I'm not sure because detailed documentation is hard to come by.

Is anyone able to advise a way forward?

Cheers, Kelli

Answer 1

[ando.kelli]

For those interested in this thread...

Apparently an older version of the pipeline had TPM as the output, but now the file .TMM.EXPR.matrix contains normalised FPKM, and is the input for the script populating the SLQ DB for Trinotateweb.

:-)

Answer 2

[egtabilin]

I think it's the opposite; the file .TMM.EXPR.matrix contains normalised TPM values as one of the intermediate files is a matrix file containing the not cross normalized TPM values (.TPM.not_cross_norm). To generate TMM normalized FPKM values, use run_TMM_normalization_write_FPKM_matrix.pl instead (I think**)