INDIANAPOLIS—A group of researchers from Indiana University School of Medicine and Indiana University Bloomington have developed a program called Multi-Omics Graph cOnvolutional NETworks (MOGONET) that integrates omics data – such as DNA, RNA and proteins – to help researchers get a comprehensive understanding of human diseases. MOGONET can identify potential biomarkers for Alzheimer’s disease and cancers from the various omics data to predict which individuals may be at higher risk.
An article about MOGONET published this month in the scientific journal, Nature Communications, includes contributions from a multidisciplinary team of researchers including senior author Kun Huang, PhD, Assistant Dean for Data Sciences and professor of biostatistics and health data sciences at IU School of Medicine; co-corresponding author Zhengming Ding, assistant professor of computer science at Tulane University; first author Tongxin Wang, a computer science graduate student at Indiana University Bloomington, and others with expertise in medicine, computer science, electrical and computer engineering, and medical and molecular genetics.
“If we put all the data together, we can learn more about the underlying biological processes and the disease outcomes,” said Huang, who is also the Director of Data Science and Informatics for the IU Grand Challenge Precision Health Initiative, a program with bold goals to develop treatments for triple negative breast cancer, as well as to prevent Alzheimer’s disease. “We can also identify potential biomarkers that can separate individuals based on their outcomes. With precision medicine, when we look at one disease, there could actually be multiple subtypes. Those subtypes may require different treatment and may have different outcomes, so we are working to identify those subtypes by looking more holistically at multi-omics data.”
Since it can be difficult to make meaningful statistical conclusions when working with multiple kinds of data, MOGONET utilizes graphs, which help show how the biomedical data are connected and interact with each other. The graph is created by using a combination of knowledge and data processing, then updated with the algorithm developed by artificial intelligence.
“The graphs are specifically useful when we are looking at the relationships of large numbers of genes, proteins or microRNAs,” said Huang, who is also a research scientist at Regenstrief Institute. “The graphs give us a way to link those molecules and put them into a network structure, which is crucial for effective deep learning. Then if we see a good performance in the prediction, we can go back to determine which biomarkers are important.”
Huang and his colleagues have demonstrated the capabilities and versatility of MOGONET through a wide range of biomedical applications, including Alzheimer’s disease patient classification, tumor grade classification in low-grade glioma (LGG), kidney cancer type classification, and breast invasive carcinoma subtype classification.
“In the future, we hope to be able to use blood or find less invasive ways to determine a patient’s risk,” said Huang, who is also the Associate Director of Data Science at IU Simon Comprehensive Cancer Center.
IU School of Medicine is the largest medical school in the U.S. and is annually ranked among the top medical schools in the nation by U.S. News & World Report. The school offers high-quality medical education, access to leading medical research and rich campus life in nine Indiana cities, including rural and urban locations consistently recognized for livability.