In other words, paraphrasing the words of Cool Hand Luke, what we’ve got here is a failure to replicate.

Results like this are remarkably common in full genome autism studies, whether (like the latest one), they’re reporting on inherited genes or (like an earlier study from the Harvard group) genetic mutations called “de novo copy number variants” (CNVs). The plot is so familiar by now that it’s worth betting on. In response to the failure of the last round of gene hunts, a new research group raises money for the largest sample of autism families ever collected. They apply the latest technology, check their results with other samples and set the computers in motion. When the results come back, lo and behold, they find that virtually all of the prior “autism genes” don’t show up significant in the latest analysis; in fact, some of the highest and best hopes of the autism genetics community are dashed by negative findings. But some brand new shiny region emerges out of the massive number crunchers and this new “discovery” goes straight to the headlines.

The sad part, of course, is that the human genome is so large is that when one runs numbers like this you’d almost always expect to find a result that approaches significance somewhere on the genome. But the gene hunters continue to sally forth undaunted. And since no one in the medical press ever bothers to go back to read any of the previous studies, they’re more than happy to file a story celebrating the triumphant march of medical science.

The funny part is, the last time we saw this dynamic, it was the CHOP group dashing the hopes of the Harvard group, which had claimed to find a “hot spot” of genetic instability on chromosome 16 (an excessive rate of deletions on chromosome 16p11). In their study published in April of this year, the CHOP group tried to replicate this finding in a follow on study of de novo CNVs. Here’s what they reported.

“We observed a similar frequency of deletions and duplications of the 16p11.2 locus in the ASD cases (~0.3%) as previously reported; however the CNV frequency in the control subjects at this locus was also comparable to that of the cases.”

In other words, the rate of mutation was no different between autism cases and controls and the Harvard group may simply not have had a large enough sample to detect such a low rate of mutations in the normal controls, where the CHOP group found plenty of deletions.

So much for the hot spot...