Artificial intelligence will improve the prognosis and treatment of autoimmune diseases

A new advanced artificial intelligence (AI) algorithm could lead to more accurate and earlier predictions, as well as the development of new treatments for autoimmune diseases, in which the immune system mistakenly attacks the body's own healthy cells and tissues. The algorithm analyzes the genetic code underlying these conditions to more accurately model how genes associated with specific autoimmune diseases are expressed and regulated, and to identify additional risk genes.

The work, developed by a team of researchers from the University of Pennsylvania College of Medicine, outperforms existing methodologies and identified 26% more new gene-trait associations, the researchers report. Their work was published today in Nature Communications.

"We all have mutations in our DNA, and we need to understand how any of these mutations can affect the expression of disease-related genes so that we can predict disease risk early on. This is especially important for autoimmune diseases," said Dajiang Liu, distinguished professor, vice chair for research and director of artificial intelligence and biomedical informatics at the University of Pennsylvania College of Medicine and co-author of the study.

“If an AI algorithm can more accurately predict disease risk, that means we can intervene earlier.”

Genetics and disease development

Genetics often underlie the development of diseases. Variations in DNA can affect gene expression, which is the process by which information in DNA is converted into functional products such as protein. How strongly or weakly a gene is expressed can influence disease risk.

Genome-wide association studies (GWAS), a popular approach in human genetics research, can identify regions of the genome associated with a particular disease or trait, but cannot pinpoint specific genes that influence disease risk. It's similar to sharing your location with a friend, but without the fine-tuning on your smartphone—the city may be obvious, but the address is hidden.

Existing methods are also limited in the detail of analysis. Gene expression may be specific to certain cell types. If the analysis does not distinguish between different cell types, the results may miss real cause-and-effect relationships between genetic variants and gene expression.

EXPRESSO method

The team's method, called EXPRESSO (EXpression PREdiction with Summary Statistics Only), uses a more advanced artificial intelligence algorithm and analyzes data from quantitative expression signatures of mononuclear cells that link genetic variants to the genes they regulate.

It also integrates 3D genomic data and epigenetics, which measures how genes can be modified by the environment to influence disease. The team applied EXPRESSO to GWAS datasets for 14 autoimmune diseases, including lupus, Crohn's disease, ulcerative colitis and rheumatoid arthritis.

"With this new method, we were able to identify many more risk genes for autoimmune diseases that truly have cell-type specific effects, meaning they only affect a certain type of cell and not others," said Bibo Jiang, assistant professor from the University of Pennsylvania College of Medicine and senior author of the study.

Potential Therapeutic Applications

The team used this information to identify potential therapeutics for autoimmune diseases. Currently, they say, there are no good long-term treatment options.

"Most treatments aim to alleviate symptoms rather than cure the disease. This is a dilemma, knowing that autoimmune diseases require long-term treatment, but existing treatments often have such bad side effects that they cannot be used long-term. However, genomics and AI offer a promising path to developing new therapeutics," said Laura Carrel, professor of biochemistry and molecular biology at the University of Pennsylvania College of Medicine and co-author of the study.

The team's work has pointed to drug compounds that can reverse gene expression in cell types associated with autoimmune disease, such as vitamin K for ulcerative colitis and metformin, which usually prescribed for type 2 diabetes, for type 1 diabetes. These drugs, already approved by the US Food and Drug Administration (FDA) as safe and effective for treating other diseases, could potentially be repurposed.

The research team is working with colleagues to test their findings in the laboratory and eventually in clinical trials.

Lida Wang, a doctoral student in the biostatistics program, and Chakrit Khunsriraksakul, who received his PhD in bioinformatics and genomics in 2022 and his medical degree in May from the University of Pennsylvania, led the study. Other authors from the University of Pennsylvania College of Medicine include Havell Marcus, who is pursuing an M.D. And medical degree; Deyi Chen, doctoral student; Fan Zhang, graduate student; and Fang Chen, postdoctoral fellow. Xiaowei Zhang, an assistant professor at the University of Texas Southwestern Medical Center, also joined the work.