I am analyzing the relevance of DNA mutations at insulated neighborhood boundaries (i.e CTCF binding sites involved in chromatin looping). Can I use MUSiC for enhancers regions and CTCF binding sites ?
MuSiC's SMG test is just, at it's heart, a test of enrichment of mutations over specific genomic intervals. If you prepare your inputs appropriately, yes, it could be used for any such test.
In that case what is the format for my ROI file?how should I supply the co-ordinates of non-coding regions?
That is entirely up to you. It might be as simple as windows tiled across the genome, or more complex by pulling in information from sources like ENCODE.