Kallisto vs Cufflinks

Question

[frida.danielsson]

Hi,

I have downloaded publicly available FPKM values for a couple of samples and I wonder what would be the biggest different between TPM values converted from these FPKMs (just by dividing by the sum of FPKMs and multiplying with 10^6) and TPM values generated with Kallisto?

Thanks for your help

[Michael]

What do you mean by biggest difference? These are different methods, so the outcome cannot be expected to correlate 100%. I'd guess that kallisto would underestimate the low and zero expressors, and shift them towards medium expressors, compared to alignment methods, by assignments to slightly ambiguous regions, where an alignment method would be more accurate in placing fragments. the best way to find out would be to just try it out.

[frida.danielsson]

Thanks for your answer. What I meant was what differences in the different methods that will have the most impact over differences between the TPM values on gene level. I assume the differences that you mention derive from the initial alignment where Kallisto's pseudo allignment will matter..?

[Istvan Albert]

What you would need to divide with is the total transcript lenght not sum of FPKMs

But just to clarify ... these are completely different concepts, neither of which is a direct measurement of gene expression hence cannot be compared directly.