Can I add TPM values of Isoforms of the same gene to get total gene expression?

Question

[baunruh]

Hello,

If I have TPM values for gene A from the UCSC database which contains multiple isoforms, and lets say I want just 1 output for gene A, can I average or add the TPM values?

If I can't do that is it possible to go back to the raw reads and add them at the raw read level?

I just want to condense genes for my dataset.

[pattymoy1990]

Hi MIchele,

I know it has been a while since you answered this question, but I was wondering if you know of any papers that have used this method so I can publish my results.

Long story short, I am working with sugarcane RNA-seq data. As I do not have a reference genome to work with, we are using a long-read Pac-Bio transcriptome (SUGIT transcriptome), this transcriptome has a high degree of redundancy/multiple isoforms (sometimes over 40 for a single gene),

Do you think your subscribed method would work/has any references backing the method? Otherwise, I could probably edit the transcriptome and manually collapse my genes of interest in the PAC-BIO transcriptome, and re-run the RNA-seq analysis, although this would be alot of effort.

Thanks

[ATpoint]

Please open a new question.

Answer 1

[Michele Busby]

If they are calculated correctly (e.g. with RSEM or similar), the quantification should be divided among the isoforms in a mutually exclusive way.

So if you have

Isoform1 3 TPM - three transcripts Isoform2 1 TPM - one transcript

The total gene should have 4 transcripts per million (i.e. 4 independent transcripts, 3 from Isoform1 and one from Isoform2.)