seanhacks/relation_prediction

NapistuDataStore Dataset

This dataset contains a complete NapistuDataStore with all artifacts published as a read-only store.

Source Data

This store was created from GCS asset: human_consensus_no_rxns (version: 20251218)

Artifacts

NapistuData (1)

• relation_prediction

VertexTensor (1)

• comprehensive_pathway_memberships

Pandas DataFrame (5)

• edge_strata_by_node_species_type
• edge_strata_by_edge_sbo_terms
• species_identifiers
• name_to_sid_map
• edge_strata_by_node_type

Usage

Load from HuggingFace Hub

The easiest way to load this dataset is using the from_huggingface class method:

from napistu_torch.napistu_data_store import NapistuDataStore
from pathlib import Path

# Load read-only store from HuggingFace Hub
store = NapistuDataStore.from_huggingface(
    repo_id="seanhacks/relation_prediction",
    store_dir=Path("./local_store"),
    revision="main"
)

# Use the store (read-only)
napistu_data = store.load_napistu_data("relation_prediction")

Configure DataConfig

You can also use this dataset in your DataConfig YAML for PyTorch Lightning experiments:

data:
  store_dir: "./local_store"
  hf_repo_id: "seanhacks/relation_prediction"
  hf_revision: "main"
  napistu_data_name: "relation_prediction"

To make the store writable (non-read-only), provide paths to the raw data files:

data:
  store_dir: "./local_store"
  hf_repo_id: "seanhacks/relation_prediction"
  hf_revision: "main"
  sbml_dfs_path: "/path/to/sbml_dfs.pkl"
  napistu_graph_path: "/path/to/napistu_graph.pkl"
  napistu_data_name: "relation_prediction"

Load Raw Data from GCS (Optional)

If you need to create new artifacts, you can convert this read-only store to a non-read-only store by loading the raw data from GCS and passing the paths directly to from_huggingface:

from napistu_torch.napistu_data_store import NapistuDataStore
from napistu.gcs.downloads import load_public_napistu_asset
from napistu.gcs.constants import GCS_SUBASSET_NAMES
from pathlib import Path
import tempfile

# Download raw data from GCS
with tempfile.TemporaryDirectory() as temp_data_dir:
    sbml_dfs_path = load_public_napistu_asset(
        "human_consensus_no_rxns",
        temp_data_dir,
        subasset=GCS_SUBASSET_NAMES.SBML_DFS,
        version="20251218",
    )
    napistu_graph_path = load_public_napistu_asset(
        "human_consensus_no_rxns",
        temp_data_dir,
        subasset=GCS_SUBASSET_NAMES.NAPISTU_GRAPH,
        version="20251218",
    )
    
    # Load and convert to non-read-only in one step
    store = NapistuDataStore.from_huggingface(
        repo_id="seanhacks/relation_prediction",
        store_dir=Path("./local_store"),
    revision="main",
        sbml_dfs_path=sbml_dfs_path,
        napistu_graph_path=napistu_graph_path,
    )
    
    # Now you can create new artifacts
    store.ensure_artifacts(["new_artifact_name"])

Links

• 🌐 Napistu
• 💻 GitHub Repository
• 📚 Napistu Wiki

Citation

If you use this dataset, please cite:

@software{napistu_torch,
  title = {Napistu-Torch: Graph Neural Networks for Biological Pathway Analysis},
  author = {Hackett, Sean R.},
  url = {https://github.com/napistu/Napistu-Torch},
  year = {2025}
}

License

MIT License - See LICENSE for details.