VSDS-vd (Virtual Screening Decoy Set for Docking)

Benchmark for comparing AI-powered and physics-based docking tools from virtual screening perspective. Evaluates 4 AI docking tools, 4 physics-based tools, and 2 AI rescoring methods. From Zhejiang University. Proposes hierarchical screening strategy.

Composite
65.8
Experimental validation
None
Stages
Hit ID
Modalities
protein_structuresmall-molecule
Task types
virtual_screeningdockingmethod_comparison
Size
description: Multiple target-specific decoy sets
License
Open
First release
2025-02
Last updated
2025-02
Official site
→ project page
Leaderboard
→ leaderboard
Dataset
→ dataset
Code / GitHub
→ repository
HuggingFace
→ HF
Paper
Benchmarking AI-powered docking methods from the perspective of virtual screening · · 2025 · 6 citations
Flags
chinese_benchmarkvirtual_screening
Experts
Groups
Hosted by
Related benchmarks
DUD-E, DEKOIS 2.0, LIT-PCBA, PoseX (Protein-Ligand Docking Benchmark)

Rubric (7-criterion)

rigor
4
coverage
3
maintenance
2
adoption
2
quality
4
accessibility
4
industry_relevance
4

Notes

Chinese research benchmark (Zhejiang University). Finds AI methods show deficiencies in physical soundness of docked structures despite good VS performance. Proposes hierarchical strategy balancing speed and accuracy.

← Back to all benchmarks

Compare:
Open comparison →