Machine intelligence-driven framework for optimized hit selection in virtual screening | JoVE Visualize

Area of Science:

Computational chemistry and cheminformatics
Drug discovery and development
Bioinformatics and computational biology

Background:

Virtual screening (VS) is crucial for identifying potential drug candidates by predicting interactions between compounds and protein targets.
Standard VS methods often suffer from high false-positive rates, necessitating improved approaches for hit identification and optimization.
Existing ligand-based and structure-based VS methods, while valuable, are often used separately, limiting their combined predictive power.

Purpose of the Study:

To introduce an advanced virtual screening framework, the automated hit identification and optimization tool (A-HIOT).
To integrate chemical and protein space analyses using machine learning for enhanced prediction accuracy in drug discovery.
To provide a robust and generalizable tool for identifying and optimizing selective drug hits for specific protein receptors.

Main Methods:

Development of A-HIOT, a framework combining chemical space-driven stacked ensemble for identification and protein space-driven deep learning for optimization.
Implementation of numerous open-source algorithms to synergistically integrate chemical and protein data.
Rigorous validation using tenfold cross-validation and independent benchmark datasets for targets like CXC chemokine receptor 4 (CXCR4) and androgen receptor (AR).

Main Results:

A-HIOT demonstrated superior performance in hit molecule identification and optimization, achieving high accuracies (e.g., 94.8% and 81.9% for CXCR4 via cross-validation).
The framework showed high accuracies on independent datasets for CXCR4 (96.2% identification, 89.9% optimization) and AR (86.8% identification, 90.2% optimization).
Results indicate A-HIOT's generalizability and robustness across different protein targets.

Conclusions:

A-HIOT effectively bridges the gap between ligand-based and structure-based virtual screening.
The framework provides a reliable approach for identifying and optimizing selective hits, significantly improving the efficiency of empirical drug discovery.
The open-source A-HIOT framework is available, promoting further research and application in the field.