A recent paper in NEJM is worth a read; it suggests a deletion on chromosome 16 predisposes to autism. The phenotype appears to be highly variable, though:

[I]n a study of the same population by investigators at deCODE Genetics, this deletion was observed at a markedly increased rate in subjects with a psychiatric or language disorder. This study showed that the deletion was present in 1 of 648 patients with schizophrenia, 1 of 420 patients with bipolar disorder, 1 of 203 patients with ADHD (the father of a child with autism, as noted above), and 1 of 3000 patients with panic disorder, anxiety, depression, or addiction. In addition, 1 of 748 patients with dyslexia carried the deletion. Overall, in the Icelandic samples, the carrier frequency among patients with autism was 1%; the frequency was approximately 0.1% among patients with a psychiatric or language disorder and 0.01% in the general population.

All these disease phenotypes can probably be thought of as extremes of some distribution of an underlying trait; I wonder how much variation in the “normal” range can be explained by de novo or rare copy number variants like this one.