99.6% is still awfully dissimilar, depending on what that number means. Thinking about single polymorphisms, of ca. 3 billion bases, that is still 120,000,000 SNPs between twins. Of course, they aren't doing full genome sequencing, and they are probably targeting loci that are know to be more divergent. Still... that similarity seems awfully low for twins. I wonder what they typically find for similarity between any two individuals. I also wonder if their lab is a little disorganized....