Control gene size for gene enrichment study

Question

[michealsmith]

I would like to test if genes containing at least one transcription factor (say MEF2A) binding sites are enriched for certain category.

I could easily come up with a TF-containing-gene list by intersecting TF binding sites bed files with gene annotation bed files, and send for enrichment study.

But question is: if one gene is big, naturally it tends to be more likely to contain TF binding sites. So should I first control gene size?

So I should normalize by assigning one parameter to each gene as: (overlap size)/(gene size) ? And then sort and select say the top 200 or 500?

[jotan]

Is the transcription factor more likely to be biologically relevant when bound to promoters? If that's true, you could just restrict the overlap to TSS+/- 1kb which would generate fragments of equal length.

[michealsmith]

No, the TF bind to everywhere, which are all biologically relevant.