Data mining for haplotypes
	represent haplotypes as regular expressions with limited gaps, e.g. (-,-,-,0,0,*,-,-) look for "overrepresented" haplotypes--occurring more than a certain # of times in the cases (note-- not a statistical comparison, but a fixed threshold) observe that any sub-haplotype of an overrepresented haplotype (introducing more *'s) is also overrepresented, and recurse to find the maximal overrepresented haplotypes