The central challenge in psychiatric drug development is causality. Epidemiological studies frequently reveal associations between a protein, a brain-imaging measure, or a lifestyle factor and the risk of a disorder, but associations alone cannot determine whether the factor drives the illness or is simply a downstream consequence of it. This ambiguity has contributed to an unusually high failure rate in late-stage psychiatric trials, where candidate drugs built on correlational evidence often fail to demonstrate benefit. Mendelian randomization addresses the problem by exploiting the random allocation of genetic variants at conception, a process analogous to the randomization in a clinical trial but conducted by nature across entire populations.

Mendelian randomization rests on a simple principle: if a genetic variant that robustly alters the level or activity of a biological molecule—such as a protein circulating in the blood or the expression of a gene in the brain—is also associated with altered risk of a disorder, this supports a causal role for that molecule in the disease process. Because genes are fixed at conception and generally not influenced by the disease itself or by environmental confounders in the same way as measured exposures, the approach can approximate the effect of a lifelong intervention. It is, in effect, a way of asking what would happen if a drug permanently modified a target, before any drug has been developed.

Yet Mendelian randomization has a well-known vulnerability: linkage disequilibrium, the phenomenon whereby neighboring genetic variants are inherited together as blocks. A variant that influences disease risk may sit near, but not within, the region controlling a candidate protein, producing a spurious causal signal. Colocalization analysis was developed to resolve exactly this problem. By examining the fine-grained pattern of genetic association in a genomic region, colocalization tests whether the same variant is driving both the protein-level signal and the disease signal. When the evidence indicates a shared causal variant, confidence that the protein genuinely participates in the disorder rises substantially, and the combination of the two methods has become a gold-standard screen for drug target prioritization.

The study brought together genome-wide association data from very large international consortia, including hundreds of thousands of participants for depression and tens of thousands for schizophrenia and bipolar disorder, alongside proteomic and transcriptomic datasets that map the abundance of thousands of proteins and genes across tissues. By systematically testing genetically predicted levels of each candidate molecule against genetic liability to each disorder, and then confirming overlaps through colocalization, the researchers were able to narrow an enormous search space of possible targets down to a short list backed by converging lines of evidence. Several of the prioritized targets encode proteins with known drugability profiles, meaning that compounds directed against them either already exist for other indications or fall within chemical classes amenable to pharmaceutical development.

Among the most striking implications of the work is the degree of shared biology it reveals across diagnostic boundaries. Depression, anxiety disorders, schizophrenia, and bipolar disorder are clinically distinct categories, but genetic studies have repeatedly shown that they overlap substantially at the level of inherited risk. The analysis reflected this reality, identifying targets whose causal signals appeared in more than one disorder. For drug developers, such pleiotropic targets carry both promise and caution: a single molecule acting on a shared pathway could potentially benefit multiple patient groups, while safety considerations become correspondingly broader, since modulating the target may influence several facets of brain function at once.

The study also underscored the importance of tissue context. Genetic variants that influence protein levels in the blood do not always do so in the brain, and psychiatric symptoms arise from neural circuitry rather than peripheral biochemistry. Where the underlying data permitted, the researchers examined whether the causal signals were consistent with expression in brain regions implicated in mood regulation, cognition, and reward processing. This layer of analysis matters for translation, because a target that appears compelling in plasma proteomics may prove irrelevant to central nervous system function, whereas one whose genetic regulation is demonstrably active in neural tissue represents a far stronger candidate for psychiatric intervention.

For patients and clinicians, the timeframe for impact should be understood realistically. Genetic target prioritization does not produce a treatment; it produces a hypothesis with unusually strong evidential support. The targets identified now require the full pipeline of experimental validation—cellular models, animal studies, medicinal chemistry, and ultimately clinical trials—before any new therapy reaches the clinic. Nevertheless, the value of the approach lies in its ability to redirect investment. History from other therapeutic areas, notably cardiology, shows that drugs developed against genetically validated targets are substantially more likely to succeed in trials than those based on other forms of evidence. Bringing the same discipline to psychiatry could measurably improve one of the least productive areas of modern pharmaceutical research.

The work also illustrates how the scale of open genetic datasets is reshaping biomedical science. The conclusions rest on the cumulative contributions of hundreds of thousands of research participants whose DNA and clinical information were aggregated across consortia worldwide, together with publicly funded resources cataloguing protein and gene regulation. No single laboratory could have assembled statistical power of this magnitude. As these datasets continue to grow, and as proteomic measurements become more comprehensive and more finely mapped, the resolution of target-screening studies of this kind will only improve, potentially extending the framework to rarer conditions, to treatment-response phenotypes, and to the prediction of side effects before trials begin.

Caution remains warranted. Mendelian randomization estimates the effect of lifelong genetic perturbation, whereas drugs act acutely and often on specific tissue compartments; developmental compensation can blunt the relevance of genetic findings; and any individual target signal requires replication in independent datasets before it can be considered settled. The colocalization framework, while powerful, depends on the density of genetic fine-mapping in each region and can be inconclusive where multiple variants contribute. Even so, the study represents a concrete advance in a field that urgently needs one: a short list of molecular targets for common mental disorders, each supported by human genetic evidence of causality, each assessed for shared mechanisms across conditions, and each grounded in data from the populations the resulting medicines would ultimately serve.

Subject of Research: Identification of therapeutic targets for common mental disorders using Mendelian randomization and colocalization

Article Title: Potential therapeutic targets for common mental disorders identified through Mendelian randomization and colocalization

Article References: Xiong, Z., Li, Z., Ji, X., Chen, H., Li, J., Peng, T., Huang, Z., Yang, L., Dong, X., Zhou, W., & Zhang, H. (2026). Potential therapeutic targets for common mental disorders identified through Mendelian randomization and colocalization. Translational Psychiatry. https://doi.org/10.1038/s41398-026-04416-5

Image Credits: AI Generated

DOI: 10.1038/s41398-026-04416-5

Keywords: Mendelian randomization, colocalization, psychiatric genetics, drug targets, depression, schizophrenia, bipolar disorder, anxiety, Translational Psychiatry, drug development, proteomics, genome-wide association studies