Question: In Enrichr: What is "Gene weight" or "levels of membership"?
1
19 months ago by
christiangriffioen10 wrote:

I have a differentially expressed gene list made with Limma. I can put just the genenames in Enrichr:

http://amp.pharm.mssm.edu/Enrichr/

But I can also add "levels of membership" / "Gene weight" which is number per gene ranging from 0 to 1. Which sounds very nice. But how do I get those numbers? I do have logFC and P.values of each gene but they ofcourse have a bigger range not a number between 0 and 1. So how do I calculate that "gene weight" here?

And why can't I find any further explanations anywhere?

I hope someone knows, thanks!

modified 19 months ago by RamRS26k • written 19 months ago by christiangriffioen10
1

Did you read the background info? http://amp.pharm.mssm.edu/Enrichr/help#background

I quote:

In general, it is recommended that non-computational users should always submit crisp sets; this is because preparing a proper fuzzy set as input requires some computational skills and submitting a fuzzy set with the data not correctly transformed can lead to erroneous results.

So their advice is to use crisp set if you don't understand how fuzzy works.

1

Because you are probably searching for the wrong terms: in the paper and documentations, the terms crisp gene set and fuzzy gene set are used. Further explanations from Enrichr Online Help:

With fuzzy set input the membership is defined as 1 for full membership, and 0 for no membership. Hence if you have a ranked list of differentially expressed genes with p-values or fold-change values, you would need to convert these to membership values between 0 and 1, keeping the ranks and scaling as appropriate.

The paper has this to say about fuzzy gene set analysis:

While intuitively fuzzy enrichment analysis should be more accurate than ‘crisp’ enrichment analysis, because ‘fuzzy’ enrichment considers the ranks and magnitude of genes in both the input set and the library sets, our initial results so far only show a marginal enhancement, utilizing the same TF-centered benchmark presented above (Figure ​(Figure1B).1B).