Gene ontology results differ from cite to software?

Question

[Merv]

Hi,

I am doing transcriptomic analysis,

When I do GO analysis in R with ClusterProfiler package I get the most enriched terms in BP as below;

• response to alcohol response to metal ion regulation of membrane potential regulation of cytosolic calcium ion concentration
  cellular divalent inorganic cation homeostasis divalent inorganic cation homeostasis proton transmembrane transport regulation of blood circulation calcium ion homeostasis cellular calcium ion homeostasis

However, when I do teh GO analysis with https://david.ncifcrf.gov/, I get different results;

• regulation of apoptotic process apoptotic process cell proliferation cellular response to peptide hormone stimulus rRNA processing ribosomal small subunit biogenesis aerobic respiration mitochondrial ATP synthesis coupled proton transport translation cytoplasmic translation

What is the difference between these two programs? Which one is correct for my gene list and why they give different results for GO analysis?

[Ram]

Please edit your post and format the list properly - the bulleting is being done on the first element only. The website's bullet buttons are not great, it's easier to manually add - before each entry.

Answer 1

[Matthias Zepper]

Yes, that is to be expected. There is a plethora of tools for GO enrichment analyses, and they use different statistic tests (DAVID e.g. Fisher's exact test) and onthologies. This paper elaborates on the differences and also has some benchmarking results.

I did not run a GO analysis in the past years, but the new GO-CAM annotations to me seem to be more powerful than those classic enrichments. Maybe you can find a user-friendly tool to screen your gene list against GO-CAMs?