A genetic perspective on ASD: What happens when 2 geeks have kids?

It's nice to see someone other than my mother and I discussing the possibility that geek + geek might = ASD.

This quote says it all.

What Dan Geschwind, director of the neurogenetics lab at UCLA, finds fascinating about dyslexia and autism is what they suggest about human intelligence: that certain kinds of excellence might require not just various modes of thinking, but different kinds of brains.

"Autism gets to fundamental issues of how we view talents and disabilities," he says. The flip side of dyslexia is enhanced abilities in math and architecture. There may be an aspect of this going on with autism and assortative mating in places like Silicon Valley. In the parents, who carry a few of the genes, they're a good thing. In the kids, who carry too many, it's very bad."
What do we make of the parents, whose difficulties make them eccentric and sometimes uncomfortable but not dysfunctional?

And what of their kids who don't inherit enough of these genes to cause full-blown classic autism or Asperger's, but are still noticeably different from the norm--the sorts of kids who get catchall labels like "Nonverbal Learning Disability" or "PDD-NOS"? The sort of kids with social, sensory, pragmatic, and attentional difficulties who also show high empathy and social cognition in some situations, and no stereotyped behavior or need for order? Is there a whole set of genes that lead to problems in non-academic areas, with fully "autistic" kids getting the whole set, dyspraxic or SPD kids getting particular subsets, and uncategorizable kids getting some odd, random mix? If so, would we see an increase in generalized nonverbal learning disabilities along with the rise in autism?