SSH2.0

SSH2.0 predicts hydrophobic interaction risk in monoclonal antibody candidates using sequence-based features to enable early biophysical assessment.

Key Features:

• SVM-based ensemble model: SSH2.0 employs an ensemble model based on support vector machines (SVM) to predict hydrophobic interactions from sequence information.
• Training dataset: The model was trained and validated on experimental data from 131 antibodies with characterized hydrophobic interaction properties.
• Feature selection via MRMD2.0 and CKSAAGP: Feature selection uses the graph-based MRMD2.0 algorithm applied to CKSAAGP-derived features to identify relevant sequence attributes.
• Performance metrics: Reported performance includes sensitivity 100.00% and accuracy 83.97%.
• Sequence-based rapid screening: Operates on sequence-derived features without requiring three-dimensional structural input, enabling rapid screening of antibody candidates.

Scientific Applications:

• Early-stage screening: Predicts hydrophobic interaction risk to screen monoclonal antibody candidates during early-stage development.
• Candidate prioritization: Enables prioritization of antibody sequences with favorable biophysical properties for further experimental characterization.
• Attrition reduction: Identifies candidates at risk of aggregation-related issues to reduce downstream attrition in antibody development.

Methodology:

Uses CKSAAGP-derived sequence features, MRMD2.0 for feature selection, and an SVM-based ensemble trained on experimental data from 131 antibodies to predict hydrophobic interactions.