Kallisto TPM values

Question

[oliveira.bhn]

Shouldn't the sum of all TPM values in the same library equals one million?

Kallisto outputs a file named "abundance.tsv", which contains tpm values for each transcript, but summing up these values won't give one million. Why is that?

Many thanks!

Ben

[erwan.scaon]

I did check some "abundance.tsv" files from many Kallisto runs in many projects. The "tpm" column always sum up to 1 million. Did you double check ?

Which Kallisto version are you using ? I did check with v0.43.0 & v0.44.0.

[oliveira.bhn]

Hi, Erwan.

Indeed I double checked and tested many libraries, but I can't get the tpm column to sum one million in any case. I am using kallisto v0.43.1.

The command line is pretty straightforward, so I have no idea what I might be doing wrong. I am just running "kallisto quant -i index -o output library_1.fq library_2.fq"

Thanks for replying.

Ben

[erwan.scaon]

You should post or link your question on the google_group.

You may also find an answer to your question from previous posts on this google_group. Does this one help ?

[oliveira.bhn]

I'll post the same question on the gougle_group then. Many thanks.

Ben.

Answer 1

[CrisMar]

Not sure if you found a solution. I had the same problem but I believe the TPMs in each sample (abundance.tsv) are not normalized. You have to normalize them when comparing them with all samples.

Normalized TPMs are in Sleuth. tpm_matrix <- sleuth_to_matrix(so, 'obs_raw', 'tpm')

https://github.com/pachterlab/sleuth/issues/165

[oliveira.bhn]

Hi, Cris.

That was so long ago, I don't actually remember how I solved it (even though I am sure I did). Still, I think you're probably right anyway.

Thanks for replying!