Dataset Viewer

The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.

YAML Metadata Warning:The task_categories "graph-machine-learning" is not in the official list: text-classification, token-classification, table-question-answering, question-answering, zero-shot-classification, translation, summarization, feature-extraction, text-generation, fill-mask, sentence-similarity, text-to-speech, text-to-audio, automatic-speech-recognition, audio-to-audio, audio-classification, audio-text-to-text, voice-activity-detection, depth-estimation, image-classification, object-detection, image-segmentation, text-to-image, image-to-text, image-to-image, image-to-video, unconditional-image-generation, video-classification, reinforcement-learning, robotics, tabular-classification, tabular-regression, tabular-to-text, table-to-text, multiple-choice, text-ranking, text-retrieval, time-series-forecasting, text-to-video, image-text-to-text, image-text-to-image, image-text-to-video, visual-question-answering, document-question-answering, zero-shot-image-classification, graph-ml, mask-generation, zero-shot-object-detection, text-to-3d, image-to-3d, image-feature-extraction, video-text-to-text, keypoint-detection, visual-document-retrieval, any-to-any, video-to-video, other

Dataset Card for SPIN-PPI

Dataset Summary

SPIN-PPI is a dataset and processed-resource collection for structure-guided protein–protein interaction (PPI) prediction with adaptive structural prior integration.

The dataset supports the manuscript:

SPIN-PPI: Adaptive structural prior integration for memory-efficient protein–protein interaction prediction

This dataset repository provides raw PPI interaction files, protein sequence dictionaries, protein structure files, and processed data used by the SPIN-PPI training and evaluation pipeline.

The repository currently supports five benchmark datasets:

  • SHS27k
  • SHS148k
  • STRING
  • SYS30k
  • SYS60k

The corresponding GitHub repository provides source code, training scripts, inference scripts, preprocessing utilities, ESMC embedding-generation scripts, and reproducibility instructions.


Dataset Description

What the dataset contains

SPIN-PPI supports multi-type PPI prediction under structure-guided and resource-aware settings. Each benchmark contains protein pairs and interaction labels. The released resources can be used to construct residue-level heterogeneous structural-prior graphs with the following relation channels:

Prior channel Description
SEQ Sequence adjacency between neighboring residues
STR-KNN Local structural neighborhoods based on Cα K-nearest neighbors
STR-DIS Distance-thresholded structural contacts
SURF Exposed-surface proximity derived from solvent-accessible surface information
LRR-REGION LRR-region connectivity derived from structure-aware LRR annotation

SPIN-PPI adaptively calibrates these structural priors and compresses residue-level features into protein-level representations for downstream GraphConv-based PPI prediction.

Supported benchmarks

Dataset Species / source context Approximate scale
SHS27k Human PPI benchmark from the SHS-style benchmark family ~27k protein-pair records
SHS148k Human PPI benchmark from the SHS-style benchmark family ~148k protein-pair records
STRING Human PPI benchmark derived from STRING-style resources Benchmark split provided in this release
SYS30k Yeast PPI benchmark constructed under the same SHS-style protocol ~30k protein-pair records
SYS60k Yeast PPI benchmark constructed under the same SHS-style protocol ~60k protein-pair records

Splits

The manuscript uses the following split strategies:

  • random
  • bfs
  • dfs

The train/validation/test ratio is 6:2:2.

BFS and DFS splits are used as harder graph-topology generalization settings than random splits.


Dataset Files

This Hugging Face dataset repository currently contains the following archives:

Archive Description Size
Sparse-SP-PPI_raw_data.zip Raw PPI interaction files and sequence dictionaries ~32.4 MB
Sparse-SP-PPI_pdb_structures.zip Protein structure files used for structural-prior graph construction ~1.82 GB
Sparse-SP-PPI_processed_data.zip Processed data used by the SPIN-PPI training and evaluation pipeline ~1.36 GB

Note: The archive names retain the legacy prefix Sparse-SP-PPI for compatibility with the original data release. They correspond to the SPIN-PPI dataset used in the manuscript.

Precomputed ESMC embeddings are not distributed as a separate archive in this Hugging Face release. Instead, ESMC-600M residue embeddings can be generated from the released protein sequences using the embedding-generation scripts provided in the GitHub repository.


Data Fields

PPI interaction files

Interaction files are tab-separated and follow the format:

protein1    protein2    interaction_type
Downloads last month
27