Much like everywhere else in the application of genomics to medicine, the hunt for the genetic origins of autism has proven harder than expected, and the genetic variants detected still only explain a minute fraction of the cases. Recent research suggests that copy number variations, many of them occurring spontaneously in the person affected, play a much bigger role than anticipated. While de novo changes are still genetic in that they reside in the genes, they also rhyme with recent suggestions that the heritability of the disorder isn't quite as high as believed.
I've written a feature article about all this which is in the latest issue of Current Biology:
Copy numbers count for autism
Current Biology, Volume 21, Issue 15, R571-R573, 9 August 2011
abstract and FREE access to pdf file