Differential Gene Expression Analysis - Rpkm Vs Readcount

Question

[Sahel]

Hi There,

This is my first time doing a differential expression analysis. I am using edgeR R package to do this analysis. So here is my question: Is it true that higher RPKM values for a gene compared to its contols means more expression of that gene?... also is it correct that if one gene is more expressed, there are more reads aligned to it (higher read count)??... So if these are correct, I can not explain the following results:

Read cound Cancer1 Normal1 Normal2 Normal3 Normal4

GeneX 2597.69 8609.76 4357.04 4406.78 4333.44

RPKM Cancer1 Normal1 Normal2 Normal3 Normal4

GeneX 62.5205 53.3113 35.4125 32.9044 32.2698

(does not read count and RPKM suggest something completely opposite about the expression of geneX?)

and this is edge R output for this gene: (edgeR runs based on read count )

Gene logConc logFC P.Value adj.P.Val

GeneX -13.0724804546548 -0.671159557171921 0.261235952608833 0.835334361204695

So would you please help me to understand what is going on? and if it is just my lack of knowledge would you please suggest me something to read? thanks alot :)

PS. If you just want to look at the expression of one gene in a set of normal and cancer samples, would you do it by comparing RPKM values, or read counts?

Answer 1

[Damian Kao]

You should read up on how RPKMs are calculated. RPKM is essentially the read count normalized by sample size AND transcript length. It can differ significantly from read counts and influence your differential expression.

Personally, I only use RPKM when I want to compare multiple genes within one sample. There really isn't much point in using RPKM when you are looking at one gene across different samples.

[Sahel]

Thanks, everything is much clear now...