NetGlycate

NetGlycate predicts glycation events on lysine residues in mammalian proteins to identify nonenzymatic modification sites formed by reducing sugars that affect protein function in contexts such as diabetes and aging.

Key Features:

• Statistical Analysis: Investigates the glycation process by analyzing the epsilon amino groups of lysine residues and identifies roles for acidic amino acids, particularly glutamate, and nearby lysine residues in facilitating glycation.
• Sequence-Based Prediction: Predicts potential glycation sites by recognizing sequence patterns where acidic amino acids are positioned C-terminally and basic lysine residues N-terminally relative to the modification site.
• Artificial Neural Networks (ANNs): Uses an ensemble of 60 artificial neural networks with a balloting procedure to enhance predictive accuracy and robustness with limited experimental data.
• Performance Metrics: Achieves a cross-validated Matthews correlation coefficient of 0.58, indicating substantial predictive capability given available datasets.

Scientific Applications:

• Diabetes research: Identifies lysine glycation sites relevant to protein dysfunction and pathology associated with diabetes.
• Aging and proteostasis studies: Maps glycation-prone lysine residues to study accumulation of glycation products during aging.
• Mechanistic and therapeutic investigations: Supports elucidation of molecular mechanisms of glycation and the development of targeted interventions and therapeutic strategies.

Methodology:

Integrates statistical analysis with machine learning on sequence-based data, employing an ensemble of 60 artificial neural networks and a balloting procedure.