The links between human diseases are more than chance. In Barcelona, researchers have built a computational framework that exposes the molecular threads binding conditions as different as asthma, breast cancer, and Huntington’s disease. By analyzing gene activity in more than 4,000 patients across 45 diseases, the Barcelona Supercomputing Center (BSC) created the most comprehensive attempt yet to explain why illnesses cluster together in people.
Published in Proceedings of the National Academy of Sciences, the study shows that nearly two-thirds of known clinical associations can be traced to shared gene expression profiles, often rooted in the immune system itself.
Tracing hidden connections
The concept of disease co-occurrence is familiar to doctors. Crohn’s disease and ulcers, for example, often appear together. But the molecular logic behind these pairings has remained obscure. Using RNA sequencing, which reveals which genes are switched on or off in a given sample, the BSC team built a “disease similarity network” that captures positive interactions (one disease heightening risk of another) and negative interactions (one disease protecting against another).
Among the striking findings: asthma and Parkinson’s disease show a positive link, while Huntington’s disease appears to guard against certain cancers. “We have known for years that patients with Huntington’s disease develop fewer solid tumours, such as lung or breast cancer, than would be expected by chance,” said Beatriz Urda, researcher at the BSC and lead author. “This study provides a possible molecular explanation for this phenomenon, revealing that many of the biological processes associated with Huntington’s disease follow pathways opposite to those of cancer.”
The immune system at the core
The analysis revealed that 95 percent of clinically related diseases share disruptions in immune pathways. That result underscores what many immunologists have long suspected: the immune system is the crossroads where unrelated illnesses collide. The team also identified potential new associations, including a link between Down syndrome and lupus, which may help clinicians refine diagnosis and anticipate complications.
Alfonso Valencia, ICREA professor and study leader, emphasized the broader implications. “This new methodology could also be particularly useful for studying rare diseases, which are often more difficult to characterise due to the scarcity of clinical data. Despite these limitations, the computational method has a capacity for detecting interactions comparable to that of more common diseases and could open the door to a better understanding of these minority pathologies,” he said.
Patients are not all the same
A critical advance of the study lies in patient stratification. By grouping individuals with the same disease into subgroups based on their molecular signatures, researchers could detect associations invisible at the population level. Some breast cancer subtypes, for instance, shared molecular ties with autism and bipolar disorder, while other subtypes displayed protective links against multiple sclerosis. These observations explain why two patients with the same diagnosis can experience divergent clinical paths.
The lesson is that disease links are not uniform. They are contingent, patient-specific, and deeply molecular. Controlled repetition becomes part of the story: the immune system again, the patient subgroup again, the repeated finding that biology resists simplification.
Opening the map
To make these insights accessible, the BSC team launched a public web platform (disease-perception.bsc.es). The resource allows interactive exploration of positive and negative disease associations and the gene expression pathways that underpin them. It is a way to move from isolated diagnoses toward a system view of human health.
The framework is not just descriptive. It suggests that clinicians could one day anticipate secondary conditions before they emerge and adapt treatments in a preventive and personalized fashion. In other words, this is not just a map of disease. It is a guidebook for medicine’s next steps.
Co-occurrence means two or more diseases appearing in the same patient more often than expected by chance. Some pairings are well known, like diabetes and cardiovascular disease. Others are less obvious, such as inverse co-occurrences where one disease seems to protect against another. By mapping gene activity, researchers can see whether co-occurring conditions share molecular pathways, like immune system alterations or disrupted cell signaling. This helps explain why certain illnesses cluster and why patient outcomes diverge, and it opens paths toward personalized prevention.
Journal: Proceedings of the National Academy of Sciences. DOI: 10.1073/pnas.2421060122
ScienceBlog.com has no paywalls, no sponsored content, and no agenda beyond getting the science right. Every story here is written to inform, not to impress an advertiser or push a point of view.
Good science journalism takes time — reading the papers, checking the claims, finding researchers who can put findings in context. We do that work because we think it matters.
If you find this site useful, consider supporting it with a donation. Even a few dollars a month helps keep the coverage independent and free for everyone.