Disco

Disco provides a cross-docking benchmark and dataset for evaluating pose prediction and ranking of ligand binding in protein-ligand complexes.

Key Features:

• Comprehensive Dataset: A cross-docking collection of 4,399 protein-ligand complexes spanning 95 distinct protein targets, annotated with difficulty levels: easy, medium, hard, and very hard.
• Benchmarking Framework: A standardized framework for evaluating pose prediction and ranking methods in molecular docking across the provided dataset.
• Customizable Dataset Generation: An integrated tool for generating customized cross-docking datasets tailored to specific experimental or benchmarking needs.

Scientific Applications:

• Algorithm Evaluation and Development: Enables assessment and comparison of docking algorithms' pose prediction and ranking performance.
• Reference Structure Selection: Supports selection of optimal docking reference structures for molecular simulations.
• Drug Discovery Research: Facilitates comparative analysis of ligand binding and docking strategies relevant to drug design.

Methodology:

Construction and distribution of a standardized cross-docking dataset categorized by difficulty (easy, medium, hard, very hard) for 4,399 protein-ligand complexes across 95 protein targets.