Dataset

Claude protein binder design (data release v1.0)

Contains 1,440 de novo miniprotein binder designs for 16 protein targets with wet-lab assay tables, raw sensorgrams, structures, and prompts.

Hugging Face ↗Aug 17, 2026

  • multimodal
  • protein-design
  • binding-kinetics
  • structure-prediction

What it is

Dataset of 1,440 Claude-designed de novo protein binders with paired wet-lab binding assays, raw sensorgrams, structures, and design provenance across 16 targets. 1,440 de novo miniprotein binders across 16 targets, with 20 parquet tables linking two-vendor SPR/BLI measurements, raw traces, structure predictions from 10 predictors, and step-level design provenance.

Key results

  • 1,440 designs of 50-120 residues
  • 16 targets; 354 binders among 1,320 assayed designs
  • 20 Parquet tables, 129,003 files, CC BY 4.0

Useful for benchmarking protein binder design systems against paired experimental binding outcomes and structural predictions.

One useful ML email a week.

Worth your time, not all of your time.

We track new ML work across papers, code, and data, then keep only the pieces worth a closer look. Selected, not collected.

Papers.Data.Code

The Weekly ML Signal

Selected, not collected.

One useful ML email a week: the papers, repos and datasets worth your attention.

Worth your time, not all of your time.