Northwestern University biophysicists have developed a new computational tool to identify gene combinations underlying complex illnesses such as diabetes, cancer, and asthma. Unlike single-gene disorders, these conditions are influenced by multiple genes working together. The number of possible gene combinations is vast, making it challenging for researchers to pinpoint the specific ones causing disease.
The new method employs a generative artificial intelligence (AI) model that amplifies limited gene expression data. This enables researchers to discern patterns of gene activity responsible for complex traits. The study will be published in the Proceedings of the National Academy of Sciences during the week of June 9.
"Many diseases are determined by a combination of genes — not just one," said Adilson Motter, the study's senior author and Charles E. and Emma H. Morrison Professor of Physics at Northwestern’s Weinberg College of Arts and Sciences. "Our model helps simplify things by identifying the key players and their collective influence."
Current methods like genome-wide association studies often fall short in detecting the collective effects of groups of genes. Motter highlighted that while identifying single genes is valuable, many observable traits result from multiple genes working together.
The research team developed an approach combining machine learning with optimization called the Transcriptome-Wide conditional Variational auto-Encoder (TWAVE). This model uses generative AI to identify patterns from limited human gene expression data, matching changes in gene expression with changes in phenotype.
"We're not looking at gene sequence but gene expression," said Thomas Wytock, a research associate involved in the study. Focusing on gene expression offers benefits such as bypassing patient privacy issues associated with genetic data and accounting for environmental factors affecting gene expression.
To demonstrate TWAVE's effectiveness, it was tested across several complex diseases and successfully identified causative genes missed by existing methods. TWAVE also revealed that different sets of genes could cause the same disease in different individuals, suggesting potential for personalized treatments tailored to specific genetic drivers.
"A disease can manifest similarly in two different individuals," Motter noted. "But there could be a different set of genes involved for each person due to genetic, environmental, and lifestyle differences."
The study received support from various institutions including the National Cancer Institute through grant P50-CA221747 and others from NSF-Simons National Institute for Theory and Mathematics in Biology and the National Science Foundation.