PANDA

PANDA predicts changes in protein binding affinity caused by mutations using protein primary sequence information rather than 3D structural data.

Key Features:

• Sequence-Based Prediction: Leverages machine learning to analyze protein primary sequence information and predict binding affinity changes without requiring 3D structural data.
• Comparative Performance: Benchmarked against structure-based methods such as MutaBind, achieving a maximum Pearson correlation coefficient of 0.52 on an external test dataset compared with MutaBind's 0.59.
• Cross-Validation Scheme: Evaluated using a cross-validation scheme that assesses generalization to mutations not present during training to address potential bias.

Scientific Applications:

• Therapeutic Development: Predicts effects of mutations on protein binding affinity to support identification of potential therapeutic targets.
• Mutagenesis Studies: Provides predictions of mutation impacts on protein–protein interactions to inform mutagenesis experiments when structural data are unavailable.

Methodology:

PANDA employs machine learning algorithms to process protein sequence data and predict changes in binding affinity upon mutation.