Magneton Data: 530,601 SwissProt Proteins with Substructure Annotations

530,601 SwissProt proteins with substructure annotations from DSSP and InterPro release 103.0, stored as sharded JSONL files. The dataset, authored by rcalef, is intended for training and evaluating substructure-aware protein representation learning models. It was last updated on Hugging Face on February 24,我们发现了一个错误。

Use Cases

• Training protein language models based on substructure annotations.
• Evaluating model performance on substructure prediction tasks.
• Benchmarking new representation learning methods against established datasets.

Strengths

• Contains 530,601 protein entries, providing a substantial scale for model training.
• Integrates annotations from two established sources: DSSP and InterPro release 103.0.
• Data is structured and stored in sharded JSONL files for efficient processing.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is known, but other metadata like file size, license, and specific column details are unavailable.
• Freshness should be verified; the last update was 2026-02-24 15:54:55.

Provenance

Source

SwissProt proteins annotated with DSSP and InterPro (release 103.0).

Collection Method

Likely compiled from public biological databases.

Time Range

Temporal coverage of the underlying protein data is not specified.

Freshness

Last updated 2026-02 我们发现了一个错误。

Geography

Spatial coverage is not applicable for this biological sequence data.