An elusive result of lipids enrichment analysis

Hi, I encountered a problem when I did enrichment analysis on MetaboAnalyst 6.0. I put in 101 lipids including 29 TAGs, but the result seemed to be a little bit weird.

The FDR of “Triradylcglycerols” set was unsignificant. But when I viewed the “Detail”, many TAGs were being enriched in the “Triradylcglycerols” set.

Could you help me to explain why? Thank you very much.

Hello! The non-significant FDR is likely due to the large total number of compounds in the Triradylcglycerols library, i.e. since the library contains >30,000 triradylcglycerol compounds, a few dozen hits in comparison may not be deemed statistically over-represented. The over representation analysis algorithm used in the enrichment module is best suited for metabolite sets with a biological context such as biological pathways, biological functions, disease states, etc. In the case of a large collection of compounds only based on chemical classification, statistical test for enrichment may not provide meaningful results. You might want to consider trying “define your own customized metabolite sets” and/or “upload a reference metabolome based on your analytical platform” options to better tailor the statistics to your research context.