The search for the genetic roots of psychiatric illnesses and behavioral disorders such as schizophrenia, autism and ADHD has a long history, but until recently, it was one marked by frustration and skepticism. In the past few years, new techniques have begun to reveal strong evidence for the role of specific genes in some cases of these conditions but in a way few people expected.

To understand what makes the new discoveries so novel, it’s necessary to appreciate how our genes can go wrong. The human genetic code can be thought of as an encyclopedia in multiple volumes. Our normal genome contains 46 chromosomes, so that’s 46 volumes. Each chromosome is a long string of the chemical DNA and the information is “written” in the form of a molecular alphabet with just four letters: A, T, C and G.

There are three ways in which something can go wrong here. First, a whole chromosome can be either missing or duplicated. This drastic change is almost always fatal. (The exceptions include Downs Syndrome.)

Second, single-nucleotide polymorphisms (SNPs, or “snips” as everyone calls them) are when a single base-pair is different, corresponding to a misprinted character.

Finally, copy-number variants (CNVs) are when a stretch of DNA is either missing (deleted), or repeated (duplicated), a bit like a page that’s either fallen out or been printed twice. As you can imagine, CNVs tend to be more serious than SNPs, because they affect more of the DNA. This is only a general rule, however. There are plenty of serious SNPs, and plenty of harmless CNVs. It all depends on where they happen, and whether they interfere with important genes.

For a long time, it was widely assumed that SNPs were responsible for psychiatric disorders, in what’s called the “common-variant model” of disease. The idea was that any given risk variant might be quite common, but it would only increase your risk of suffering a disease by a small amount. Those who carried a large number of risk variants would develop the disease. Those with a moderate number might get mild symptoms, and so on.

Yet this just didn’t work out. About ten years ago, it became feasible to scan huge numbers of SNPs quickly and cheaply. These “genome-wide association studies” (GWAS) tested hundreds of thousands of variants. Moving quickly to exploit the new technology, psychiatrists conducted GWAS after GWAS comparing people with diseases to those without — but very little came out. There are a few common SNPs which seem to be associated with some disorders, like autism and schizophrenia, but only a handful, and they have very small effects.

The same is true of other areas of genetics as well. Known SNPs only account for a small percentage of the risk of many common medical disorders or traits that are thought to be genetic, like heart disease, height, and obesity. Psychiatry, however, is especially barren.

It’s always possible that even bigger studies, looking at even more SNPs, might be able to find more associations. Some recent research has suggested that there are many variants of extremely small effect still to be found for schizophrenia and bipolar disorder. So there’s still life in SNPs, but it’s fair to say that compared to 5 years ago people regard them with rather more skepticism.

CNVs, however, have just taken off — in the nick of time, some say. What’s emerging from these studies, however, may be, in its own way, revolutionary as well.

Psychiatric interest in CNVs was sparked by a landmark paper that appeared in Nature in September 2008. It was authored by an international consortium of schizophrenia researchers, led by employees at an Icelandic company, deCODE Genetics. They found a number of CNVs which seemed to be associated with schizophrenia.

Since then, CNV studies have taken off in the same way as GWAS did 5 years before. There’s now good evidence for the involvement of deletions and duplications in autism, attention-deficit hyperactivity disorder (ADHD), schizophrenia, and intellectual disability (aka mental retardation). By contrast, however, studies in bipolar disorder have been negative.

A typical finding is that, comparing a group of patients to some healthy controls, the rate of CNVs in the patients is higher, and these CNVs are especially likely to disrupt genes involved in brain development and function. For instance, in one recent ADHD study from a group led by Cardiff University’s Nigel Williams and colleagues, published in The Lancet, the authors found large, rare deletions or duplications in 15 percent of the children with the disorder compared to 7 percent of the controls.

When this study appeared last year, many media sources reported it as evidence that scientists had found “the ADHD gene”. There are two problems with this interpretation. Firstly, only a small proportion of patients carried “large rare” CNV. 85% did not carry any, although more detailed future studies, able to detect smaller CNVs, might have found more (the smaller they are, the harder they are to detect.)

The deeper problem is that there wasn’t just one CNV. In fact, there were dozens of different large deletions or duplications in the ADHD group. This is similar to the results of other CNV studies.

Furthermore, even when the same CNV turns up repeatedly in many independent patients, these patients very often have different diseases. Many of the leading risk CNVs for autism, say, have also been found in ADHD and schizophrenia, epilepsy, or intellectual disability – and vice versa.

To take just one example, the “15q13.3” deletion, so called because it affects a particular part of Chromosome 15, has been found in people with schizophrenia, epilepsy, autism, and possibly even antisocial behaviors.

So although scientists set out trying to discover the genetic causes of named psychiatric disorders like “autism” and “schizophrenia,” they’re increasingly finding that these diagnoses don’t correspond to particular genes at all.

Instead, it may be that these diagnostic categories are just describing particular symptoms of certain genetic disorders. So, rather than saying that 15q13.3 deletion causes schizophrenia, for example, in the future we might say that some of the features of schizophrenia are amongst the effects of the 15q13.3-deletion-syndrome.

It’s only early days yet, but as this research advances further, and as technology allows ever-smaller CNVs to be picked up, these kinds of genetic findings may present a serious challenge for existing psychiatric diagnostic systems.

Are you a scientist? And have you recently read a peer-reviewed paper that you would like to write about? Please send suggestions to Mind Matters editor Gareth Cook, a Pulitzer prize-winning journalist at the Boston Globe. He can be reached at garethideas AT or Twitter @garethideas.