PON-Sol

PON-Sol predicts the effects of amino acid substitutions on protein solubility to assess impacts relevant to protein expression, stability, aggregation, and crystallizability.

Key Features:

• Dataset Utilization: Trained on the largest available dataset of experimentally verified solubility-affecting amino acid substitutions collected from literature.
• Predictive Accuracy: Reports a normalized correct prediction ratio of 0.491 in cross-validation and an independent test set accuracy of 0.432 and classifies variants as solubility-increasing, solubility-decreasing, or neutral.
• Comparative Performance: Demonstrates superior predictive performance relative to other predictors focused on protein solubility and aggregation.
• Exhaustive Variant Scanning: Enables evaluation of all possible amino acid substitutions within a protein for their effects on solubility.

Scientific Applications:

• Disease Research: Predicts the impact of disease-related amino acid substitutions to aid understanding of pathogenic mechanisms linked to protein aggregation and misfolding.
• Protein Engineering: Supports exploration of substitutions to design proteins with enhanced solubility or improved crystallizability for biotechnological applications.
• Recombinant Protein Expression: Informs optimization of heterologous recombinant protein expression by predicting substitution effects on solubility.

Methodology:

PON-Sol employs machine learning trained on a large-scale, experimentally verified dataset of solubility-affecting amino acid substitutions to predict increases, decreases, or neutral effects on protein solubility.