iThermo

iThermo predicts thermophilic proteins from protein sequence information to identify thermostable proteins for biotechnological and industrial applications.

Key Features:

• Input data: Uses protein sequence information as the basis for prediction.
• Multi-layer Perceptron classifier: Employs a multi-layer perceptron (MLP) trained to distinguish thermophilic and non-thermophilic proteins.
• Multi-feature fusion strategy: Integrates multiple sequence-derived features to improve predictive performance.
• Performance: Reports 96.26% accuracy on an independent test set.

Scientific Applications:

• Enzyme technology: Identifying thermophilic enzymes for high-temperature industrial catalysis.
• Protein engineering: Selecting thermostable proteins and scaffolds for engineering and stabilization efforts.
• Biofuel production: Discovering thermotolerant proteins for biomass conversion processes.
• Pharmaceuticals: Identifying thermostable proteins relevant to formulation and processing under elevated temperatures.

Methodology:

Model training used a benchmark dataset of 1,368 thermophilic proteins and 1,443 non-thermophilic proteins, applying a multi-layer perceptron classifier with a multi-feature fusion strategy for training and validation.