AI Model Predicts Future SARS-CoV-2 Mutations Using 'Genetic Dialect'

AI Model Predicts SARS-CoV-2 Mutations

Researchers from Florida Atlantic University's College of Engineering and Computer Science have developed an innovative artificial intelligence model capable of predicting future mutations in the SARS-CoV-2 virus. This breakthrough comes five years after COVID-19 was declared a global pandemic, as the virus transitions to endemic status 1

1

Deep Novel Mutation Search (DNMS)

The new method, called Deep Novel Mutation Search (DNMS), utilizes deep neural networks to predict mutations in protein sequences. Focusing on the SARS-CoV-2 spike protein, responsible for the virus's entry into human cells, the researchers employed a protein language model to forecast potential new mutations 2

2

ProtBERT: Understanding the Virus's 'Dialect'

At the core of DNMS is ProtBERT, a language model fine-tuned to comprehend the "dialect" of SARS-CoV-2 spike proteins. This model evaluates potential mutations based on several factors:

1. Grammaticality: How likely or "correct" a mutation is according to learned rules.
2. Semantic change: The similarity between the mutated sequence and the original protein.
3. Attention change: A novel measure applied to mutation prediction 3

   3

   .

Parent-Child Mutation Prediction Model

Unlike previous research that typically examines changes to a reference protein sequence, DNMS introduces a parent-child mutation prediction model. This approach uses an existing protein sequence (parent) to predict mutations and analyzes how these mutations might evolve over time 1

1

Predicting Viral Fitness

The study also explored the relationship between predicted mutations and viral fitness. Findings indicate that mutations with high grammaticality, small semantic change, and low attention change were associated with higher viral fitness. This suggests that mutations adhering to the protein's biological "rules" and causing minimal disruption are more likely to benefit the virus 2

2

Related Stories

AI Model EVE-Vax Predicts Viral Mutations to Enhance Future Vaccine Development

AI-Driven Platform Optimizes Antibodies to Combat Evolving SARS-CoV-2 Variants

AI-Powered PandemicLLM Revolutionizes Infectious Disease Forecasting

AI Advances Promise to Transform Pandemic Preparedness and Response

Implications for Public Health

Dr. Xingquan "Hill" Zhu, senior author and professor at FAU's Department of Electrical Engineering and Computer Science, emphasized the model's ability to rank all possible mutations and identify those most likely to occur in the future. This capability could prove invaluable for guiding experimental research and helping public health officials prepare for new mutations before they spread widely 3

3

Advantages Over Traditional Methods

DNMS offers several advantages over traditional wet-lab experiments, which can be costly and time-consuming. By using sequence data alone and leveraging the biological rules that proteins follow, DNMS provides a more efficient and cost-effective method for predicting viral mutations 1

1

The research, sponsored by the United States National Science Foundation, represents a significant step forward in our ability to anticipate and prepare for future viral threats. As SARS-CoV-2 continues to evolve, tools like DNMS may prove crucial in staying ahead of the virus and protecting public health.