CSP-Analyzer

CSP-Analyzer automates detection and classification of chemical shift perturbations (CSPs) in two-dimensional HSQC NMR spectra to identify ligand binding events for fragment-based drug discovery.

Key Features:

• Automated Analysis: Evaluates large collections of 2D HSQC spectra to detect and quantify chemical shift perturbations across samples.
• Machine Learning Integration: Uses a machine-learning classifier with a SMOTE-ENN (Synthetic Minority Over-sampling Technique - Edited Nearest Neighbors) statistical discrimination step to classify and assess ligand binding events.
• Bias Reduction: Automation of CSP detection and classification reduces user-introduced variability in spectrum assessment.
• Efficiency: Enables rapid evaluation and binning of spectra to support high-throughput fragment screening workflows.

Scientific Applications:

• Fragment-based drug discovery: Automated identification of CSPs across 2D HSQC spectra to detect ligand–protein interactions and prioritize fragments for follow-up and lead optimization.

Methodology:

Processes 2D HSQC NMR data with a Python-based machine-learning classifier trained to recognize CSP patterns associated with ligand binding and employs SMOTE-ENN for statistical discrimination and class imbalance handling to classify and bin spectra by binding status.