protein solubility

protein solubility predicts continuous solubility values from amino acid sequences using machine learning regression to inform recombinant protein engineering and expression studies.

Key Features:

• Continuous Solubility Prediction: Provides continuous numerical solubility predictions from amino acid sequences rather than binary soluble/insoluble outputs.
• Machine Learning Algorithms: Integrates machine learning regression methods including support vector machines (SVM) and reports an R² of 0.4115 for SVM predictions.
• Indirect Activity Prediction: Leverages correlations between protein activity and solubility to enable indirect inference of activity from sequence-derived solubility predictions.
• Experimental Design Guidance: Produces continuous solubility values to prioritize proteins and guide experimental designs aimed at improving solubility.
• Open-Source Workflow: Provides an open-source machine learning workflow as IPython notebooks on GitHub (https://github.com/xiaomizhou616/protein_solubility).

Scientific Applications:

• Protein Engineering: Supports engineering of proteins with improved solubility to enhance biocatalyst efficiency.
• Biotechnology Research: Aids experimental design and hypothesis generation in studies of protein expression and activity.
• Data Analysis Template: Serves as a workflow template for analyzing expression and solubility datasets.

Methodology:

Machine learning regression models, including support vector machines (SVM), applied to amino acid sequence-derived features; workflow implemented as IPython notebooks (https://github.com/xiaomizhou616/protein_solubility).