TSLAmy: A Novel Amyloid Hexapeptide Aggregation Prediction Approach Based on Two-Stage Learning

Area of Science:

• Biochemistry
• Computational Biology
• Drug Discovery

Background:

• Identifying aggregation-prone peptides is crucial for understanding amyloid-related diseases and developing new peptide-based drugs.
• Current methods for identifying such peptides can be costly and time-consuming.

Purpose of the Study:

• To develop TSLAmy, a computational model for predicting amyloid hexapeptides.
• To improve the efficiency and reduce the cost of identifying aggregation-prone peptides.

Main Methods:

• A two-stage learning framework was employed, involving feature extraction and a prediction model.
• Autoencoders were used for balanced dataset partitioning based on sequence and physicochemical features.
• Physicochemical and ESM-2 sequence features were extracted, followed by aggregation prediction.

Main Results:

• TSLAmy achieved a prediction accuracy of 0.8493, outperforming existing state-of-the-art methods.
• The model successfully predicted the aggregation potential of all 64,000,000 possible hexapeptides.
• Analysis identified specific amino acids contributing to aggregation-prone hexapeptides.

Conclusions:

• TSLAmy offers a powerful computational approach for identifying aggregation-prone hexapeptides.
• The findings provide valuable insights for advancing amyloidosis research and peptide drug development.