Nobody has done an extensive evaluation of competing methods. Choose a recently published method, such as
Enriched Domain Detector, which is described in a
journal article. It doesn't make sense to do motif finding for broad regions. That's for sharp peaks, such as when the experiment is about transcription factors.