Molecular HIV Inhibition Prediction Dataset

Serving as adapted from MoleculeNet for molecular property prediction, specifically to predict whether molecules inhibit HIV. It is part of the Open Graph Benchmark and is intended for a binary classification task evaluated with ROC-AUC.

Use Cases

• Train graph neural networks to predict HIV inhibition labels from molecular graph structures.
• Benchmark binary classification models for molecular property prediction tasks using ROC-AUC.
• Analyze molecular features within the Open Graph Benchmark framework for HIV-related property inference.

Strengths

• Dataset is part of the standardized Open Graph Benchmark, facilitating model comparison.
• Task is clearly defined as binary classification for molecular HIV inhibition prediction.
• Evaluation uses the specific metric of ROC-AUC, providing a consistent performance measure.

Limitations

• Dataset is described as 'small', which may limit model training and generalization.
• Specific details on the number of molecular graphs, features, or class balance are not provided.
• Original source and data collection methodology are not detailed in the provided input.

Provenance

Source

Adapted from MoleculeNet by teams at Stanford for the Open Graph Benchmark (OGB).

Collection Method

Adapted from an existing molecular property dataset.

Freshness

Last updated on 2023-02-07.