ANuPP

ANuPP predicts amyloid-fibril-forming peptides and aggregation-nucleating regions in peptide and protein sequences using atomic-level features and an ensemble-classifier framework.

Key Features:

• Prediction targets: Identifies amyloid-fibril-forming peptides and aggregation-prone regions (APRs) within protein sequences.
• Atomic-Level Feature Analysis: Uses atomic-level characteristics, including spatial arrangement and chemical properties of atoms, to characterize aggregation nucleation.
• Ensemble-Classifier Framework: Employs an ensemble-classifier that integrates multiple predictive models to produce aggregated predictions.
• Predictive performance on hexapeptides: Validated on 1,279 hexapeptides with 10-fold cross-validation yielding AUC 0.831 and 77% accuracy, and on a blind test of 142 hexapeptides with AUC 0.883 and 83% accuracy.
• Protein-Level Analysis: Detects APR regions within proteins with an average SOV of 48.7% across 37 proteins.

Scientific Applications:

• Neurodegenerative disease research: Enables identification of aggregation-prone regions relevant to diseases such as Alzheimer's, Parkinson's, and prion disorders.
• Drug discovery and development: Supports targeting of specific aggregation-prone regions for therapeutic intervention against amyloid-related conditions.
• Protein engineering: Guides modification of protein sequences to reduce aggregation propensity and improve stability.

Methodology:

Analysis of atomic-level features (including spatial arrangement and chemical properties of atoms) combined with prediction by an ensemble-classifier integrating multiple predictive models.