Tamchemist2026/openbind-blind-challenge-1

OpenBind Blind Challenge 1 — Test Set

Target: Zika virus (ZIKV) NS2B-NS3 protease Challenge: OpenBind Blind Challenge 1, run by OpenBind in partnership with OpenADMET and the ASAP Discovery Consortium Submission portal: huggingface.co/spaces/openadmet/openbind-blind-challenge-1

Dataset Summary

This dataset provides the compound list for the first OpenBind Blind Challenge: 400 small molecules, ranging from fragments to lead-like compounds, each with an experimentally determined (but withheld) bound structure against the Zika virus NS2B-NS3 protease. Participants are given each compound's SMILES string and asked to predict the 3D structure of its complex with the protease — the reference (experimental) structures are not released until after the challenge closes.

Dataset Structure

Data Instances

A single CSV file, openbind-blind-challenge-1-TEST-BLINDED.csv, with one row per compound:

Molecule_Name,SMILES
blind_challenge_001,O=C(Nc1ccc2c(c1)CNC2)c1cccc2cn[nH]c12
blind_challenge_002,O=C(Nc1ccc2c(c1)CNC2)c1cccn2nccc12
...

Data Fields

Field	Type	Description
Molecule_Name	string	Compound identifier, blind_challenge_001 through blind_challenge_400. Use this exact identifier to name your predicted structure files (e.g. blind_challenge_001.pdb) when submitting.
SMILES	string	The compound's structure as a SMILES string.

Supporting Files

Participants also need, from the challenge tutorial repository:

• The NS2B-NS3 construct's amino acid sequence (FASTA)
• A representative apo (ligand-free) structure of the optimised NS2B-NS3 system, PDB code 8PN6
• Baseline notebooks and a submission-validation script

The closed conformation NS2B-NS3 construct used throughout this challenge is deposited under PDB code 29HN.

Dataset Creation

Source Data

Compounds were soaked into ZIKV NS2B-NS3 crystals and structures determined by X-ray crystallography. The compound set combines:

• Fragments from the EUbOpen DSiP extension and the Enamine Essential Fragments Libraries
• Follow-on compounds designed by the ASAP Discovery Consortium during the hit-to-lead phase of their Zika NS2B-NS3 discovery programme
• Fragment analogues and follow-up compounds designed by OpenBind

Diffraction data were collected via unattended data collection at beamlines I03 and I04-1, Diamond Light Source, UK (proposal lb42888), automatically processed with in-house pipelines, with automatic binding-event detection and ligand fitting via autoPanDDA and Pipedream, and ligand-binding sites manually checked and refined in XChemExplorer.

Acknowledgements

OpenBind received funding from the UK Department of Science, Innovation and Technology under grant number G2-SCH-2025-06-16537. Some data featured in this challenge was generated by ASAP Discovery under NIH NIAID Antiviral Drug Discovery (AViDD) grant U19AI171399. Many compounds are derived from the medicinal chemistry campaign conducted by the ASAP Discovery Consortium, disclosed in a related preprint.

The OpenADMET team thanks Radial (part of the Astera Institute) for continuing support of these blind challenge efforts.

Contact

Questions, support, and announcements: join the #openbind-challenge channel on Discord.