ProG-SOL: Predicting Protein Solubility Using Protein Embeddings and Dual-Graph Convolutional Networks.

Area of Science:

• Biophysics
• Protein Engineering
• Computational Biology

Background:

• Protein solubility is a critical biophysical property for biochemical engineering applications.
• Existing protein solubility prediction methods struggle with generalization and regression tasks.
• Improved prediction accuracy is needed for effective protein engineering.

Purpose of the Study:

• To develop an advanced method for predicting protein solubility.
• To enhance the generalization performance of solubility prediction models.
• To improve solubility prediction for both classification and regression.

Main Methods:

• Developed ProG-SOL, a sequence-based dual-graph convolutional network.
• Utilized both protein pretrained graphs and protein evolutionary graphs.
• Evaluated the model on independent test sets for classification and regression.

Main Results:

• ProG-SOL demonstrated superior performance compared to existing methods.
• Achieved improved classification and regression results on independent test sets.
• The model framework shows potential for predicting other protein properties.

Conclusions:

• ProG-SOL offers a significant advancement in protein solubility prediction.
• The dual-graph convolutional network approach enhances accuracy and generalization.
• The method has broad applicability in protein engineering and related fields.