QSAR Bioconcentration Classes: 779 Chemicals with 9 Dragon Descriptors

779 manually-curated chemicals with experimental bioconcentration factor (BCF) and mechanistic class labels. The data includes 9 Dragon molecular descriptors, CAS numbers, and SMILES strings, and was used to develop QSAR classification trees. The dataset was created by Grisoni et al. and published in 2016.

Use Cases

• Training classification models to predict a chemical's bioconcentration mechanism based on molecular descriptors.
• Developing regression models to estimate the Bioconcentration Factor (BCF) from structural features.
• Comparing the performance of CART and Genetic Algorithm techniques on a curated environmental dataset.
• Analyzing the relationship between specific Dragon descriptors (e.g., MLOGP, nHM) and bioaccumulation behavior.

Strengths

• Contains 779 manually-curated chemical records with experimental BCF values.
• Includes 9 specific Dragon molecular descriptors for model development.
• Provides a predefined training/test split (584/195 compounds) preserving class proportions.

Limitations

• Row count for the full dataset is not explicitly stated in the provided metadata.
• Column-level documentation is absent; field semantics must be inferred from the description or original paper.
• Last update date is unknown; freshness unverified.

Provenance

Source

Original manuscript by Grisoni et al., 2016, Environment International.

Collection Method

Manually-curated BCF data for chemicals, with descriptors calculated using Dragon software.