Hi all,

I'm having difficulty to understand Pvalue that DEXSeq outputs and hope someone can help me.

I know that DEXSeq allows testing among multiple conditions and it tests if any affects exon usage across all conditions.

In my Experiment I have three conditions: Control, L and LL.  I expect to see three fold change values and three Pvalues for each pair of conditions, If DEXSeq tests for all conditions.

However, I can only find two fold change values: logFC_C_L and logFC_C_LL and one adj.Pvalue. 

So my questions are how do I understand the Pvalue, is it selected from Min( P<c,l>; P<c,ll>; P<l,ll>) where P represents Pvalue? And, why logFC_L_LL is missing?

Thank you in advance.

Jason

Hi Jason,

DEXSeq tests for differences in exon usage between different conditions. This means that if you have three different conditions, DEXSeq will give you one p-value for each exon, where the null model would be "the different conditions do not have an effect on exon usage". You get a p-value for each exon, where the significant p-values indicate that you are rejecting the p-value: at least one of the conditions has an effect on exon usage, such that the exon usage is different with respect to the other conditions. 

The fold changes in DEXSeq are precisely comparing different conditions and this is also the reason why you are finding only two fold changes, since one of the conditions is taken as the "denominator" for the fold change.

Alejandro