A question about adjust-pvalue in gene expression analysis

Question

[Shamim Sarhadi]

Hi I am doing a meta-analysis with some publicly available datasets, I want to finding DEG's between case and control in my datasets seperately and then combine the p-valu's , but in some of my dataset with adjust-pvalue less than 0.05 , there is not any DEGs , now I think I have two options, 1- Increasing adjust p-value(less restrictive condition) in those dataset that have not DEG's with adjust pvalue less than 0.05, 2- Excluding those datasets that have not DEG's with this adj-pvalue.

any idea? Thanks in advance

[Farbod]

Dear Shamim, Hi.

If the "there is not any DEG" is true, so it is what it is.

why you are searching for any DEGs in a condition they are not exist? Maybe you know the answer before starting the analysis or there is some strong biological concept you are sure about ?

The cutoff of adjusted p-value and FDR depends on many things. In my idea you can mention them as some observations with no DEGs but do not relax your cutoff. It is not wise to adjust your FDR and FC cutoffs based on attempting particular genes to pass this threshold.

Also, please have a look at this interesting post.

Hope that helps.

~ Best

[Shamim Sarhadi]

Hi Dear Farbod, yes and I corrected my mistake Thanks for your explanation, I want to compare two biological states that I think they are very near to each other, and I want to find DEG's that contribute to biological mechanisms that led to transition between these two states, according your comment it's better to mention them as some observations with no DEGs ,Thanks again for useful links All the best

[Farbod]

Shamim Jan, Thank you.

Would you also have a look at this post that I have called it "To DEG or not to DEG", please.

~ Best

Answer 1

[Farbod]

Hi,

And please have a look at this paper.