Instead the GWAS, which operate without specific underlying biological models, which differentiates them from the candidate gene approach, have brought up risk loci for psychiatric diseases about whose function little, if anything, was known. Although the effects of the newly identified genome-wide supported risk variants SCH772984 chemical structure on protein structure and function will ultimately only be answered by molecular and cell biology, neuroimaging can provide global measures of the pathways involved (Inkster et al., 2010). For example, several SNPs on genes coding for subunits of the

GABA-A receptor have been associated with bipolar disorder, but they do not alter protein sequence and their function is unknown (Craddock et al., 2010). One possibility is that they affect the expression levels of subunits

and thus the functionality of the resulting GABA receptors, which could be tested with noninvasive markers of the GABA system derived from MEG and MRS (Muthukumaraswamy et al., 2009). This rationale is even more relevant for the rare CNVs that are enriched in autism, mental retardation, ADHD, schizophrenia, and possibly other mental disorders (Owen et al., 2010 and Williams et al., 2010). These CNVs are only present in a small minority of cases and their immediate molecular effects therefore unlikely to explain the pathophysiological pathways underlying the disorder. However, many CNVs on different genes may affect the same functional system—for example NMDA receptor complexes in the case of schizophrenia (Kirov et al., 2011)—and thus converge on intermediate PD-1/PD-L1 inhibitor 2 (neurobiological) phenotypes that can be mapped out with noninvasive techniques (Figure 2). One example is the mismatch negativity, a negative deflection of the event-related potential that is evoked by deviants in trains of auditory stimuli and can be modulated

by antagonists of the NMDA receptor (Stephan et al., 2006). Such a noninvasive index of a putative pathological first process might then become a proof-of-concept marker for intervention studies. The choice of biological target for potential gene effects can be informed by endophenotypes, which are heritable traits that can be defined at any level from biochemistry to behavior. In order to meet the definition of an endophenotype for a particular disease (Gottesman and Gould, 2003), a trait needs to occur in both patients and unaffected relatives more frequently than in the general population. The crucial idea is that endophenotypes are more closely associated with specific genes than the clinical phenotype and can aid in the discovery of new disease genes or in the definition of genetic subtypes of the disease. Several such endophenotypes have been proposed on the basis of neuroimaging findings, for example, reduced GABA concentrations as measured by MRS in relation to major depression (Hasler and Northoff, 2011).