Dear community,

I want to combine proteomic and transcriptomic data for my project.

Since there are only 40 proteins measured in the assay I am wondering if I can just extract the normalized counts from my DESeq2 object and apply a Wilcox rank sum test or if it is recommended to extract LFCs and p_values/adjusted p_values after DESeq2 procedures (from negativ binomial model).

Thank you for your assessments.

The problem with the Wilcox test is that it has lower power at low sample size. That is where the likes of DESeq2 (and other tools) come in -- borrowing information from all genes to better estimate the overall mean-variance trend. What you get from these tools is typically more reliable that simple tests such as the Wilcox unless your sample size is really large. I would (without more context given) use the output of DESeq2.