Protein Graph Dataset for Enzyme Classification

PROTEINS is a molecular property prediction dataset for binary classification, aiming to predict whether molecules are enzymes. The dataset is designed for use with a 10-fold cross-validation protocol and was created by the graphs-datasets organization. It was last updated on February 7, 2023.

Use Cases

• Predict enzyme classification from graph node and edge features using graph neural networks.
• Benchmark binary classification model accuracy on molecular graph data via 10-fold cross-validation.
• Train models to learn molecular representations from graph structures for property prediction.

Strengths

• Dataset is structured for a standardized 10-fold cross-validation evaluation.
• Designed for a specific binary classification task (enzyme vs. non-enzyme).

Limitations

• Exact sample size, number of features, and class balance are unknown.
• Dataset freshness is unclear as the last known update was in early 2023.

Provenance

Source

graphs-datasets organization on Hugging Face.

Collection Method

null

Time Range

null

Freshness

Last updated 2023-02-07.

Geography

null