Compound-Protein Affinity Predictions for 81 Reproductive Toxicity Targets

Yu Ma's dataset contains predicted compound binding affinities for 81 reproductive toxicity targets, generated using a classification pretraining-regression fine-tuning framework. The data is stored in an 89.3 KB XLSX file and was last updated on April 15, 2026. The framework was evaluated across six dual-encoder models and three data-split strategies, with the best-performing model achieving an R² of 0.797 and an MSE of 0.370.

Use Cases

• Distinguishing toxicity effects based on predicted target affinity spectra.
• Identifying broad-spectrum and target-specific structural alerts for reproductive toxicity.
• Revealing key protein targets for six sex-specific reproductive diseases.
• Training or benchmarking affinity prediction models for novel chemical scaffolds.

Strengths

• Model framework demonstrated an average R² improvement of 0.324 under a stringent cluster-split scenario.
• Predictions cover 81 reproductive toxicity targets.
• Best-performing model achieved an R² of 0.797 and an MSE of 0.370.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.

Provenance

Source

figshare

Collection Method

Generated via a machine learning framework for target-based affinity prediction.

Time Range

null

Freshness

Last updated 2026-04-15 06:44:15; freshness should be verified.

Geography

null