Multi-target QSAR Models for Class I HDAC Inhibitors with 1,215 Compounds

1,215 compounds from the ChEMBL database were used to construct multi-target QSAR models for predicting inhibitory activity against class I HDAC isoforms. The models, including linear and non-linear classifiers, were built using 13 deviation descriptors and achieved accuracies exceeding 90% on sub-training, test, and validation sets. Author G.G. Tu published the dataset on figshare in May 2026, which includes results from virtual screening, drug-likeness assessment, and molecular dynamics simulations.

Use Cases

• Training machine learning models for compound activity prediction based on the described 13 deviation descriptors.
• Virtual screening of compound libraries like ZINC to identify potential HDAC inhibitors.
• Assessing drug-likeness of candidate molecules using in silico tools like SwissADME.
• Studying protein-ligand interactions through docking and molecular dynamics simulations.

Strengths

• Models were validated on sub-training, test, and validation sets with reported accuracies exceeding 90%.
• Dataset is based on 1,215 compounds sourced from the authoritative ChEMBL database.
• Analysis includes multiple modeling approaches (linear and six non-linear models) and subsequent molecular dynamics validation.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the underlying data tables is unknown, which may limit suitability assessment.
• Last updated 2026-05-28 15:03:06; freshness should be verified.

Provenance

Source

Compounds obtained from the ChEMBL database.

Collection Method

Models constructed using the Box–Jenkins moving average method with 13 deviation descriptors.

Freshness

Last updated 2026-05-28 15:03:06.