Using extended-connectivity fingerprints with Laplacian-modified Bayesian analysis in high-throughput screening...

Area of Science:

• Computational chemistry
• Drug discovery
• Biochemistry

Background:

• Escherichia coli dihydrofolate reductase is a key target for antimicrobial drugs.
• Predicting drug efficacy computationally is crucial for efficient drug discovery.
• High-throughput screening generates large datasets for computational analysis.

Purpose of the Study:

• To evaluate the efficacy of 2D computational methods for predicting inhibitors of E. coli dihydrofolate reductase.
• To compare the performance of 2D methods against potentially more complex 3D approaches.
• To assess the computational cost-effectiveness of different predictive modeling strategies.

Main Methods:

• Utilized extended-connectivity fingerprints (ECFPs) for molecular representation.
• Applied Laplacian-modified Bayesian analysis for predictive modeling.
• Tested the methods on a large dataset of 50,000 compounds provided by McMaster University's High-Throughput Screening Lab.

Main Results:

• 2D methods, ECFPs combined with Laplacian-modified Bayesian analysis, achieved competitive results in predicting enzyme inhibitors.
• The 2D approach demonstrated surprisingly strong performance despite the problem's apparent suitability for 3D methods.
• The computational cost associated with the 2D methods was notably low.

Conclusions:

• 2D computational approaches are viable and cost-effective for identifying inhibitors of E. coli dihydrofolate reductase.
• ECFPs and Laplacian-modified Bayesian analysis offer a practical alternative to 3D methods in this context.
• This study highlights the potential of simplified computational strategies in drug discovery pipelines.