DUBS: A Framework for Developing Directory of Useful Benchmarking Sets for Virtual Screening

Area of Science:

• Computational Chemistry
• Structural Biology
• Drug Discovery

Background:

• Virtual screening methods are crucial for drug discovery but lack standardized benchmarking data.
• Inconsistent formats for protein and ligand structures hinder reliable evaluation of virtual screening tools.

Purpose of the Study:

• To introduce the Directory of Useful Benchmarking Sets (DUBS) framework for creating standardized virtual screening benchmarking sets.
• To provide a flexible and efficient tool for generating benchmark datasets from the Protein Data Bank.

Main Methods:

• DUBS utilizes a simple text-based input format and the Lemon data mining framework.
• It efficiently accesses, organizes, and formats data from the Protein Data Bank for virtual screening software.
• The framework is implemented as a Python script for user modification and extensibility.

Main Results:

• DUBS can create a benchmark dataset in under 2 minutes.
• It offers a standardized representation for benchmarking data, addressing a critical issue in the field.
• The framework allows users to define custom benchmarking datasets.

Conclusions:

• DUBS provides a valuable community resource for standardizing virtual screening benchmarking.
• The framework enhances the reliability and efficiency of evaluating drug discovery methods.
• Its simple, flexible design facilitates widespread adoption and customization.