Pxr Challenge Cofolding: Multi-method Predictions for hPXR with 600+ Ligands

Over 40,000 cofolding predictions generated by author dargason using three methods (OpenFold3, Boltz2, Chai1) for the nuclear hormone receptor hPXR. The ligand set contains more than 600 molecules, including known binders, agonists, and inactives, sourced from the OpenADMET PXR Challenge. The dataset was last updated on 2026-05-18.

Use Cases

• Benchmark cofolding method performance based on predictions from three distinct algorithms.
• Train or validate machine learning models for binding affinity prediction based on the multi-method prediction data.
• Analyze structural interactions for known PXR binders and agonists versus inactives based on the described ligand categories.
• Investigate the impact of untemplated, default-settings runs on prediction outcomes for a specific protein target.

Strengths

• Contains over 40,000 individual cofolding predictions.
• Includes predictions from three distinct computational methods (OpenFold3, Boltz2, Chai1).
• Ligand set of over 600 molecules provides known binders, agonists, and inactives for comparative analysis.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Freshness should be verified as the last update was 2026-05-18.

Provenance

Source

Ligands derived from the OpenADMET PXR Challenge; predictions by dargason.

Collection Method

Cofolding predictions run using three methods (OpenFold3, Boltz2, Chai1) with default settings and untemplated.

Freshness

Last updated 2026-05-18 01:47:36