IPs-GRUAtt

IPs-GRUAtt predicts phosphorylation sites in host proteins during SARS-CoV-2 infection to elucidate virus–host phosphorylation-mediated molecular mechanisms.

Key Features:

• Attention-based bidirectional GRU network: Employs bidirectional gated recurrent units (GRUs) combined with an attention mechanism to capture dependencies and contextual information from protein sequences.
• Phosphorylation site prediction: Identifies phosphorylation sites in proteins affected by SARS-CoV-2 infection.
• Attention-derived motif recognition: Uses the attention mechanism to highlight relevant sequence motifs and patterns associated with phosphorylation.
• Comparative performance: Comparative analyses demonstrated outperforming state-of-the-art machine-learning methods and existing models in predicting phosphorylation sites.

Scientific Applications:

• Mapping infection-induced phosphorylation: Locates phosphorylation sites to analyze post-translational modification changes in SARS-CoV-2–infected host cells.
• Elucidating virus–host interactions: Supports identification of host signaling pathways manipulated by SARS-CoV-2 through phosphorylation changes.
• Informing therapeutic strategies: Provides phosphorylation site information relevant for prioritizing targets for interventions against COVID-19.

Methodology:

The model is a bidirectional GRU network enhanced with an attention mechanism trained on protein sequence data from SARS-CoV-2–infected host cells to predict phosphorylation sites.