Multi-Scale Representation Learning on Proteins: Protein-Ligand and Enzyme Structures

Protein-ligand structures from PDBBind (v2019) and protein structures from an enzyme dataset are provided. The data supports the paper 'Multi-Scale Representation Learning on Proteins' by Vignesh Ram Somnath of ETH Zurich. It is used for tasks like ligand binding affinity regression and protein function prediction.

Use Cases

• Predict ligand binding affinity based on multi-scale protein graph representations.
• Classify protein function using learned multi-scale structural and sequence embeddings.
• Benchmark graph encoders on protein-ligand interaction tasks.
• Explore memory-efficient protein surface segmentation via molecular superpixels.

Strengths

• Data is sourced from established databases PDBBind (v2019) and a published enzyme dataset.
• Associated model reportedly outperforms baselines on most dataset splits for regression and uses 10x fewer parameters for classification.

Limitations

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

Provenance

Source

PDBBind (v2019) and an enzyme dataset as described in the paper.

Collection Method

Collected from public structural biology databases.