DeepIPs

DeepIps predicts phosphorylation sites in host proteins during SARS-CoV-2 infection to identify altered signaling pathways and molecular mechanisms.

Key Features:

• Targeted prediction: Predicts phosphorylation sites specifically in host proteins from SARS-CoV-2 infected cells.
• Deep learning architecture: Integrates a word embedding method with a convolutional neural network (CNN) and a long short-term memory (LSTM) network for sequence modeling.
• Spatial and temporal pattern capture: Uses CNNs to model local (spatial) sequence motifs and LSTMs to model long-range (temporal) dependencies in protein sequences.
• Performance validation: Demonstrates improved prediction performance over general phosphorylation predictors as assessed by independent testing.

Scientific Applications:

• Pathway mapping: Identifies phosphorylation sites to map signaling pathways altered by SARS-CoV-2 infection.
• Host–virus interaction studies: Supports analysis of molecular mechanisms of SARS-CoV-2 modulation of host proteins via phosphorylation.
• Target prioritization: Informs prioritization of phosphorylation-mediated therapeutic targets in virology and drug discovery research.
• Cross-disciplinary use: Applicable to research in virology, immunology, and related biomedical fields studying post-translational modifications.

Methodology:

The model encodes protein sequences using a word embedding method and applies a CNN-LSTM architecture to capture spatial (CNN) and temporal (LSTM) patterns, with performance evaluated by independent testing.