An advanced artificial intelligence (AI) algorithm has been developed to more accurately model the expression and regulation of genes associated with specific autoimmune diseases, as well as to identify additional genes at risk. This method has shown better performance than existing methodologies and has successfully identified 26% more novel gene and trait associations. This breakthrough may lead to improved predictions and new treatments for autoimmune diseases, which occur when the immune system mistakenly attacks the body’s own healthy cells and tissues. The algorithm delves into the genetic code underlying these conditions.
The team at Penn State College of Medicine has developed new methods to better understand how genes related to autoimmune diseases are expressed and regulated, as well as to identify additional high-risk genes.
According to the researchers, their work surpasses current methodologies and has discovered 26% more new gene and trait associations. Their findings were published in Nature Communications on May 20.
“We all have some DNA mutations, and it’s important to determine how these mutations can affect gene expression associated with disease in order to predict disease risk early on. This is significant in advancing our understanding of autoimmune diseases,” the researchers explained.
AI algorithms can play a crucial role in predicting disease risk, especially for autoimmune diseases,” said Dajiang Liu, a distinguished professor and director of artificial intelligence and biomedical informatics at the Penn State College of Medicine. “Early interventions can be carried out if AI algorithms are able to accurately predict disease risk.”
Genetic variations can influence disease development by affecting gene expression, which is the process of converting DNA information into functional products such as proteins. The level of gene expression can impact the risk of developing a disease.
Genomics play a significant role in understanding how genetic variations contribute to disease development.Genome-wide association studies (GWAS) are a popular method in human genetics research that can identify areas of the genome linked to a specific disease or characteristic. However, they are unable to identify the exact genes that influence disease risks. This can be compared to sharing your location with a friend while having the precise location setting turned off on your smartphone. The general area may be clear, but the specific address is hidden. Current methods also have limitations in the level of detail in their analysis. Gene expression can be specific to certain types of cells. If the analysis does not differentiate between different cell types, it may miss genuine causal relationships between them.The research team has developed a method called EXPRESSO (EXpression PREdiction with Summary Statistics Only) to study genetic variants and gene expression. This method uses a more advanced artificial intelligence algorithm to analyze data from single-cell expression quantitative trait loci, which connects genetic variants to the genes they regulate. In addition, it incorporates 3D genomic data and epigenetics to assess how genes may be modified by the environment to impact disease. The team applied EXPRESSO to GWAS datasets for 14 autoimmune diseases, such as lupus, Crohn’s disease, ulcerative colitis, and rheumatoid arthritis.
According to Bibo Jiang, a study by the Penn State College of Medicine has discovered more genes associated with autoimmune disease using a new method. These risk genes have cell-type specific effects, meaning they only impact a certain type of cell and not others.
Using this new information, the team hopes to identify potential therapeutics for autoimmune disease. Currently, there are limited long-term treatment options available for these diseases.
Most current treatments only aim to relieve symptoms, rather than curing the disease. This presents a challenge for individuals with autoimmune diseases.
Laura Carrel, a biochemistry and molecular biology professor at the Penn State College of Medicine and co-senior author of the study, stated that current treatments for autoimmune diseases often have severe side effects that limit their long-term use. However, she also mentioned that genomics and AI show potential for developing new therapeutics for these conditions. The team’s research identified drug compounds that could reverse gene expression in cell types associated with autoimmune diseases, such as vitamin K for ulcerative colitis and metformin for type 1 diabetes, which is typically prescribed for type 2 diabetes. These drugs are already approved for use.The Food and Drug Administration has approved certain drugs for treating specific diseases, but researchers are now exploring the possibility of repurposing these drugs for other conditions. The research team is collaborating with others to confirm their findings in a laboratory and eventually in clinical trials. Lida Wang, a doctoral student in biostatistics, and Chachrit Khunsriraksakul, who obtained a doctorate in bioinformatics and geonomics in 2022 and a medical degree from Penn State in May, co-led the study. Other authors from Penn State College of Medicine include: Havell Markus, currently pursuing a doctorate and a medical degree, and Dieyi Chen, a doctoral candidate.; The research paper was contributed to by Lida Wang, Chachrit Khunsriraksakul, Havell Markus, Dieyi Chen, Fan Zhang, Fang Chen, Xiaowei Zhan, Laura Carrel, Dajiang. J. Liu, and Bibo Jiang.
The work was supported by funding from the National Institutes of Health (grant numbers R01HG011035, R01AI174108, and R01ES036042) as well as the Artificial Intelligence and Biomedical Informatics pilot grant from the Penn State College of Medicine.
