Nine Compound Activity Classes and Descriptors for Inverse QSAR Analysis

Nine compound activity classes were assembled from the ChEMBL version 22 database. The dataset also includes a descriptor list for use in quantitative structure-activity relationship modeling. It was created by Tomoyuki Miyao at the University of Bonn for research into differential evolution algorithms.

Use Cases

• Benchmarking differential evolution algorithms for inverse QSAR based on the provided descriptor list.
• Training molecular activity prediction models using the nine assembled compound classes.
• Developing novel molecular descriptors by analyzing the provided reference descriptor set.

Strengths

• Data is sourced from ChEMBL version 22, a well-known, curated bioactivity database.
• Includes a descriptor list from a published research paper (F1000Research 2017), providing a methodological foundation.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for large-scale modeling.
• Last update date is unknown; freshness unverified.

Provenance

Source

ChEMBL version 22 database.

Collection Method

Assembled as described in the referenced research paper.