I performed and IPA analysis on differentially expressed genes. It will give me log10p and ratio on y-axis vs list of pathways on X-axis. My question is is there any rule by which we can say one pathway is most enriched or not. Like in terms of the p-value of number of genes in the pathway.
There is no one "right" rule, but a rule that is typically used is adjusted p-value < 0.05.
The p-value should be adjusted for multiple comparisons (usually via false discovery rate [FDR] control). Don't base your 0.05 (or whatever) cutoff on unadjusted p-values please.
Edit: If you mean which pathway is MOST enriched, usually softwares provide some sort of enrichment score and you can choose the pathway with the highest enrichment score. If you're only giving the p-values, then a heuristic would just be choosing the pathway with the smallest p-value.
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
Does that mean the lowest FDR p-value (highest-log10p bar) will be considered most significantly enriched? Can we quantitate level of significant enrichment based on level of FDR p-value? I believe the FDR p-value will suggest all significantly enriched pathways.
Edited my post above. Ideally you'd have some form of enrichment score. (The reason I somewhat dislike picking the smallest p-value is because you should already set a p-value cutoff beforehand, and select your hits based off that. I see p-values to be useful in hypothesis testing: answering whether there is sufficient evidence to think that an effect exists. I don't necessarily like using them in quantifying effect sizes.)
Keep in mind: Why do you want to pick out the MOST enriched pathway? The MOST enriched pathway versus the second or third most enriched pathway? Do you think some simple pathway analysis is going to tell you that the most enriched pathway is more biologically relevant than the second or third most enriched pathway?