TEMPO

TEMPO predicts site-specific mutations in SARS-CoV-2 and other infectious viruses using a transformer-based model informed by phylogenetic tree-based sampling to anticipate viral evolutionary changes.

Key Features:

• Phylogenetic tree-based sampling: Uses a phylogenetic tree-based sampling method to generate sequence evolution data that captures evolutionary relationships among virus sequences.
• Transformer-based architecture: Employs a transformer model to learn high-level representations of sequence data for precise site-specific mutation prediction.
• Comprehensive dataset evaluation: Validated on extensive SARS-CoV-2 datasets and benchmarked against several state-of-the-art baseline methods, demonstrating superior mutation prediction performance.
• Cross-virus applicability: Applied to additional infectious viruses to demonstrate robustness and adaptability beyond SARS-CoV-2.

Scientific Applications:

• Epidemiology and public health: Predicts viral mutations to inform vaccine development and deployment strategies and to anticipate changes that could affect treatment efficacy or transmission dynamics.

Methodology:

Generates sequence evolution data using a phylogenetic tree-based sampling method, and processes that data with a transformer model to predict potential mutation sites in viral genomes.