Immune-Related DNA Methylation Signatures, Age Prediction, And Cross-Tissue Assessment Via Biologically Informed Neural Networks
Our bodies are incredibly complex, and scientists are constantly looking for new ways to understand how we age and what makes us susceptible to certain diseases. One exciting area of research involves “DNA methylation,” which is like a set of chemical tags on our DNA that can turn genes on or off, influencing how our cells behave. These tags change over time, and some of these changes are linked to aging and health.
Recently, a study used a sophisticated type of artificial intelligence, called “biologically informed neural networks,” to explore these DNA methylation patterns, specifically focusing on those found in our immune system. Think of these neural networks as smart computer programs that learn from vast amounts of biological data, helping us find hidden connections.
The researchers discovered specific immune-related DNA methylation patterns that are very good at predicting a person’s age, even across different types of tissues like blood from children and cells from infant cheeks. The AI model was remarkably accurate, predicting age with an average error of less than two years. This consistency across different tissues is a significant step forward, suggesting these patterns are robust indicators.
Beyond age prediction, the study also uncovered intriguing links between these methylation patterns and health. For instance, specific methylation changes in genes called IL6 and CXCR4 were found to be associated with important “perinatal outcomes,” which are health events occurring around the time of birth, such as preterm birth and lower birth weight. It’s important to note that while these associations were identified, it doesn’t necessarily mean the methylation changes directly cause these outcomes, but they do highlight potential biological pathways for further investigation.
This work provides a powerful new framework for understanding how our immune system’s epigenetic landscape changes with age and how these changes might contribute to health and disease. By making these complex biological signals more interpretable, this research opens doors for developing new ways to monitor health and potentially intervene earlier.
Source: link to paper