MPSC

MPSC predicts protein cleavage sites and integrates UVR and ROS susceptibility assessments to evaluate protein degradation in human skin proteomes.

Key Features:

• Bidirectional Recurrent Neural Network (BRNN): Utilizes a BRNN architecture to predict cleavage sites for nine matrix metalloproteinases (MMPs), four cathepsins, elastase-2, and granzyme-B, with validation against simple and complex protein mixtures and reported superior performance to existing models.
• Integration of UVR/ROS susceptibility: Combines protease cleavage site predictions with ultraviolet radiation (UVR) and reactive oxygen species (ROS) susceptibility assessments based on amino acid composition to evaluate environmental and endogenous degradative forces.
• Experimental validation: Predictive models were tested against experimental datasets including exposure of extracellular matrix (ECM) proteins and complex cell-derived proteomes to MMP-9.

Scientific Applications:

• Proteome analysis in skin aging: Enables analysis of susceptibility of dermal extracellular matrix (ECM) proteins to photo-ageing by predicting UVR, ROS, and proteolytic vulnerabilities.
• Biomarker identification: Supports identification of protein biomarkers associated with tissue damage and age-related changes by highlighting proteins with elevated degradation susceptibility.
• Protease degradomics: Facilitates identification of proteins targeted by specific proteases, aiding discovery of proteolytic substrates and potential therapeutic targets.

Methodology:

Models employ a Bidirectional Recurrent Neural Network (BRNN) and integrate amino acid composition-based UVR/ROS susceptibility predictions; model performance was evaluated against experimental datasets including ECM and complex cell-derived proteomes exposed to MMP-9.