I suggest to follow the Dr. Vindhya Mohindra's recommendation.
There are many methods to detect signatures of selection using SNP, AFLP, SSR… However, most of them are based on comparisons among populations. In this special case: more pop under selection can be better. If you have enough information about your population, two populations can work for your purpose.
Which kind of species are you using (natural population, crop, plant, animal)? Probably you can find some supporting data in databases.
Basically these methods detect changes in allelic frequencies, haplotype diversity and differences in LD pattern in at least two populations. Based on these values we can make inferences as to how likely a given locus is a candidate to be under selection.
If you have any questions, feel free to ask below. You made a very interesting question and it is aways important to all of us to have this kind of discussion.