Hugging Face's logo

Datasets:
pvd232
/
HopTF

License:
Dataset card Files
xet
 Community
Dataset Viewer
The dataset could not be loaded because the splits use different data file formats, which is not supported. Read more about the splits configuration. Click for more details.
Couldn't infer the same data file format for all splits. Got {NamedSplit('train'): ('json', {}), NamedSplit('validation'): ('csv', {}), NamedSplit('test'): ('json', {})}
Error code:   FileFormatMismatchBetweenSplitsError

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

HopTF Artifacts

This dataset stores HopTF data artifacts, contracts, and diagnostics.

Code lives on GitHub: https://github.com/cpudan/HopTF/tree/peter

Layout 

contracts/                  Contract documents and artifact registry snapshots
raw_sources/                Raw source artifacts and source indexes
processed/                  Processed TFAtlas, embedding, H5AD, and token artifacts
experiments/                Experiment diagnostics, logs, and run records
promoted/                   Promoted reusable outputs
manifests/20260420/         Migration manifests and file-level provenance

Primary Artifacts

  • processed/protein_embeddings/tf_atlas_morf_isoforms_esmc_600m/: ESM-C 600M MORF isoform embedding matrix, mask, metadata, vocab, sequences, source workbook, and reproduction notes.
  • raw_sources/joung_tfatlas/published_h5ad/GSE217460_210322_TFAtlas_subsample_raw_csr.h5ad: single sparse CSR TFAtlas H5AD with X = log1p_cp10k and layers["counts"] = raw UMI counts.
  • processed/borzoi_tokens/chr1_chr11_chr17_chr19_chr22_524kb_full/: tokenized Borzoi locus panel for full 524288 bp windows on chr1, chr11, chr17, chr19, and chr22.
  • processed/tf_atlas_h5ad/final_210322/: final TFAtlas H5AD shard surface.
  • processed/tf_atlas_zarr_csr/: pre-sorted CSR raw-count cache.
  • processed/tf_atlas_raw_direct_esmc_600m/: raw-direct TFAtlas ESM-C 600M comparison embedding artifact.
  • raw_sources/protein_models/esmc/: ESM-C 300M and 600M local model snapshots.
  • raw_sources/joung_tfatlas/: retained Joung TFAtlas raw and published source artifacts.
  • experiments/: diagnostics and logs from linear-probe and AlphaGenome/Hopfield runs.

Key Files 

processed/protein_embeddings/tf_atlas_morf_isoforms_esmc_600m/tf_atlas_morf_isoforms_esmc_600m_mean_non_special.npy
processed/protein_embeddings/tf_atlas_morf_isoforms_esmc_600m/tf_atlas_morf_isoforms_esmc_600m_mean_non_special_mask.npy
processed/protein_embeddings/tf_atlas_morf_isoforms_esmc_600m/tf_atlas_morf_isoforms_esmc_600m_mean_non_special.meta.json
raw_sources/joung_tfatlas/published_h5ad/GSE217460_210322_TFAtlas_subsample_raw_csr.h5ad
raw_sources/protein_models/esmc/esmc-600m-2024-12/data/weights/esmc_600m_2024_12_v0.pth
raw_sources/protein_models/esmc/esmc-300m-2024-12/data/weights/esmc_300m_2024_12_v0.pth

Notes

  • ESM-C is the active protein embedding family in this dataset.
  • The older ESM-3 artifact and snapshot were removed.
  • The Borzoi token panel is a locus library, not a pooled summary vector.
  • The single CSR H5AD is the simplest paper-aligned TFAtlas runtime surface.

Provenance

Migration and restoration records are stored under:

manifests/20260420/HF_CURRENT_FILE_MANIFEST.json
manifests/20260420/HF_ARTIFACT_LAYOUT_PLAN.json
manifests/20260420/HF_FINAL_LAYOUT_VERIFICATION.json
manifests/20260420/HF_FINAL_MAPPING_COVERAGE.json
manifests/20260420/RESTORED_PRE_MIGRATION_ARTIFACTS.json
manifests/20260420/HF_RESTORED_REQUIRED_ARTIFACTS_VERIFICATION.json
Downloads last month
22
Total file size:
201 GB