EPSOL

EPSOL predicts protein solubility in Escherichia coli expression systems using deep learning to support recombinant protein production and manufacturability decisions.

Key Features:

• Novel Deep Learning Architecture: EPSOL employs a deep learning architecture that leverages multidimensional embedding to automatically extract comprehensive feature representations from protein sequences.
• Performance Metrics: EPSOL reports an accuracy of 0.79 and a Matthews correlation coefficient of 0.58 on benchmark evaluations.
• Large-Scale Screening Capability: EPSOL enables large-scale screening of sequence variants to identify candidates with enhanced manufacturability for recombinant protein production.

Scientific Applications:

• Protein Engineering: Predicts solubility to guide design of proteins with improved expression characteristics in Escherichia coli, reducing experimental iterations.
• Manufacturing Cost Reduction: Forecasts solubility to prioritize constructs less prone to insoluble aggregates and thereby reduce production costs.
• Prediction for Novel Recombinant Proteins: Provides solubility predictions for new recombinant proteins expressed in Escherichia coli to support research and development decisions.

Methodology:

EPSOL uses a deep learning architecture integrating multidimensional embedding to automatically generate feature representations from protein sequences.