OGB ogbg-ppa: Protein Association Graph Classification (Open Graph Benchmark)

Provider
Open Graph Benchmark team (Hu et al., Stanford)
License
CC0-1.0 (source)
Size
5 files
Format
JSON Lines (in sampled files)
Stored at
Hugging Face (https://huggingface.co/datasets/OGB/ogbg-ppa)
Access
Free and public. No account with the data provider is needed.
Category
Machine Learning & AI
Source
https://arxiv.org/abs/2005.00687

About OGB ogbg-ppa: Protein Association Graph Classification (Open Graph Benchmark)

ogbg-ppa from the Open Graph Benchmark (OGB), the standard suite of large, realistic graph machine learning benchmarks with fixed splits and leaderboards. Each graph is a protein-protein association neighbourhood extracted from the protein association networks of 1,581 species covering 37 broad taxonomic groups; neighbourhoods are undirected 2-hop subgraphs of fewer than 300 nodes, nodes are proteins, and edges are biologically meaningful associations carrying 7-dimensional features giving the approximate confidence of each association type. The task is graph classification: predict which of 37 taxonomic groups a network comes from, evaluated by accuracy. This Hugging Face copy stores the graphs as JSON Lines (train.jsonl about 9.6 GB, valid.jsonl about 5.5 GB, test.jsonl about 4.2 GB, measured 2026-11-14) with edge index, edge attributes, node count and label per graph, and the card shows how to convert rows to PyTorch Geometric Data objects. Uses include benchmarking graph neural networks, graph pooling and readout methods and scalable graph learning. Load with Hugging Face datasets or read the JSONL with pandas, then build graphs with PyTorch Geometric or DGL; the official ogb Python package also downloads it. OGB lists the ogbg-ppa licence as CC-0; cite Hu et al., Open Graph Benchmark (NeurIPS 2020).

How to load this dataset in Python

from datasets import load_dataset
ds = load_dataset("OGB/ogbg-ppa")
print(ds)

Data structure

File types in sampled files: JSON Lines

Files

Example file: test.jsonl

Analyze this dataset without downloading it

On Scigantic, OGB ogbg-ppa: Protein Association Graph Classification (Open Graph Benchmark) opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.

This page is also available as Markdown and as JSON.

Related Machine Learning & AI datasets