When someone is diagnosed with depression and anxiety, it rarely surprises their clinician. When someone with schizophrenia has mood symptoms, it rarely surprises their family. Comorbidity has always been psychiatry’s everyday reality. What has been harder to explain is why these overlaps are so common and whether the diagnostic boundaries in our manuals reflect distinct biological conditions or a set of human-made labels drawn around a shared landscape.
A new paper in Nature takes a major step toward answering that question. By analyzing common genetic variation across 14 psychiatric disorders, researchers created one of the most detailed maps yet of what is shared, what is distinct, and where today’s categories may be closer cousins than we think.
One Million Lives, Fourteen Diagnoses, One Underlying Problem
The researchers drew on genome-wide association data from 14 disorders, including schizophrenia, bipolar disorder, major depression, posttraumatic stress disorder (PTSD), anxiety disorders, attention-deficit/hyperactivity disorder (ADHD), autism, anorexia nervosa, obsessive-compulsive disorder (OCD), Tourette syndrome, and several substance use conditions. Altogether, the dataset included more than 1,056,000 cases, making it one of the largest cross-disorder genetic efforts ever assembled.
The disorders group into a smaller number of underlying genetic dimensions. Five broad genetic factors, together, explained most of the genetic variance across the disorders, around two-thirds on average. These five groupings looked roughly like this:
A schizophrenia-bipolar cluster, reflecting how strongly those two conditions share genetic risk
An internalizing cluster, centered on major depression, PTSD, and anxiety
A neurodevelopmental cluster, linking autism and ADHD (with some overlap from Tourette’s)
A compulsive-related cluster, anchored by anorexia and OCD (again with some overlap)
A substance use cluster, spanning opioid, alcohol, and cannabis-related problems, and nicotine dependence
If you have ever wondered why certain diagnoses so often travel together, this is the genetic version of the answer: The risk is not evenly shared across everything, but it is shared in structured neighborhoods.
Blended Clusters
The most striking overlaps were within two families: the schizophrenia-bipolar factor and the internalizing factor. The disorders inside these clusters were more difficult to separate into neat biological boxes.
This does not mean schizophrenia and bipolar disorder are the same, or that anxiety and depression are interchangeable. People’s lived experiences can be profoundly different. But it does suggest that the underlying inherited liability, as captured by common variants, is heavily shared, which helps explain why symptoms and diagnoses blur in practice.
The team also tested a general psychopathology factor, sometimes called a p-factor. This is a broad vulnerability that raises risk across many forms of mental illness, rather than steering someone toward one specific diagnosis. They did find evidence for such a general factor, but they also found reasons to be cautious about treating it as the main story. Many genetic signals did not fit neatly into a single “general risk” explanation, implying that a layered model works better: There may be a broad foundation, but above it sit more specific domains that matter for understanding real differences.
Disorder Hotspots in Our Genes
Not all genetic sharing is spread evenly across the genome. Some regions look like crossroads where multiple conditions intersect. These hotspots are important because they move the conversation from “Everything overlaps” to “Overlap happens a lot here.” That is a more actionable scientific clue.
The most widely shared risk across all 14 disorders was enriched for broad biological processes, such as the machinery that helps control when and how genes are turned on and off. It is too early to translate this into treatments, but this is the kind of signal researchers look for when trying to connect genetic risk to brain mechanisms.
If you have struggled with mental health, it can be tempting to read genetics as fate. This paper does not support that conclusion. Genetic risk is only one piece of the puzzle, and the authors note major limitations, including the fact that most analyses were restricted to European-ancestry datasets.
The more pressing implication is this: Our diagnostic labels may be less like separate diseases and more like regions on a shared map, with fuzzy borders and frequent overlap. That framing can reduce stigma. It can also validate what many people already know from experience: Symptoms do not always respect the categories we use to describe them.
If future psychiatry is guided not just by symptoms but by patterns of underlying biology, what changes? One possibility is that treatment could become less siloed. Instead of asking, “Which diagnosis is this?” clinicians might also ask, “Which domain of risk is most active here?” such as internalizing distress, neurodevelopmental vulnerability, or substance-related compulsivity. Another possibility is that research trials could be designed around these shared domains, aiming for therapies that help common comorbid presentations rather than one narrow label.
For now, this study offers evidence that the overlaps people live with are not just clinical messiness. They are, at least in part, written into the shared architecture of genetic risk.