Mapping Complex Diseases
A computer model of epidemiological data from 1.5 million people illuminates the genetic origins of many common diseases.
Researchers at Columbia University have mapped the overlap between 161 different diseases by studying epidemiological data from 1.5 million patients. Among their findings is a strong overlap between schizophrenia, bipolar disorder, and autism, suggesting that these three diseases may be caused by a shared group of genes. The researchers hope others will use their map to further investigate the genetic bases of the diseases they studied–genetics that in most cases are poorly understood.
Certain diseases caused by single genetic mutations are correlated with other conditions in well-known ways, says Andrey Rzhetsky, the leader of the mapping project, who is now a professor of genetic medicine at the University of Chicago. For example, the same mutation in the gene for hemoglobin, the protein that carries oxygen in the blood, causes sickle-cell anemia but protects against malaria. Unlike sickle-cell anemia, however, most diseases aren’t caused by a single mutation. The genetic factors underlying most common diseases, such as diabetes, addiction, and heart disease, are complex and poorly understood. But Rzhetsky found connections between genetically complex diseases, too.
Using health records from the Columbia University Medical Center, Rzhetsky’s group examined the likelihood that a patient with one genetically complex disease–for example, diabetes–also had one of the 160 other diseases under study, such as an autoimmune disorder. The researchers concluded that certain groups of genes can predispose a person to multiple diseases, while others can predispose a person to one disease while protecting against another.
Rzhetsky’s group did not look at gene expression or DNA sequence data in any of the patients, so the study provides no specific evidence that diseases with a tendency to occur together share common genetic risk factors. (A person with a psychiatric illness might develop diabetes because of poor eating habits, not because the same genes cause both diseases.) But in general, says Rzhetsky, the researchers can infer that the disease correlations are caused by shared genes rather than environmental effects, because their sample size was so large and the correlations were so strong.
“This [work] opens the door to a new way of looking at disease associations,” says Peter Karp, director of the bioinformatics research group at SRI International. Rzhetsky’s project is one of only a few that combine bioinformatics with medical informatics, Karp says. Bioinformatics uses computing to work on molecular-biology problems like analyzing gene expression, while medical informatics uses computing to process patient records. Taking this unusual approach, says Karp, the Columbia researchers “discovered new disease correlations from first principles.”
Some of the disease correlations found in the mapping exercise have been previously observed, but many are new. Though Rzhetsky cautions that the work is preliminary, the results give some support to theories that autoimmune disorders and bacterial or viral infections might put people at risk for autism and diabetes.
The strongest results concern neurological disorders. “We’re the first to quantify the overlap between schizophrenia, bipolar, and autism,” says Rzhetsky. (See the slide show for maps showing the overlap between these diseases, as well as a map of migraine.)
For many of the diseases the Columbia researchers examined, specific genetic risk factors are known. Researchers can examine the role of these genes in more poorly understood diseases that overlap with a better understood one.
Karp says that he hopes Rzhetsky’s work will “lead to new avenues of research” and that other researchers will expand on it to include more patients, increasing its statistical power.
AI is here.
Own what happens next at EmTech Digital 2019.