PaRSnIP

PaRSnIP predicts protein solubility from amino acid sequences using a gradient boosting machine model trained on sequence- and structure-derived features.

Key Features:

• Gradient Boosting Machine Prediction: Applies a gradient boosting machine algorithm to integrate sequence-based and approximated structural features for solubility classification.
• Feature Importance Analysis: Reports importance scores for all training features, highlighting determinants such as fractions of exposed residues and histidine-rich tripeptide stretches associated with solubility outcomes.

Scientific Applications:

• Protein Engineering and Biomanufacturing: Screens protein variants for improved solubility to optimize recombinant protein expression and production workflows.

Methodology:

PaRSnIP extracts sequence and predicted structural features from protein sequences, trains a gradient boosting machine model, evaluates performance using independent test sets, and computes feature importance to interpret solubility determinants.