pinder

What it is and what it does

Pinder is a dataset and resource for protein-protein docking research, providing access to a large collection of protein structures and interaction data hosted on Google Cloud Storage. It includes monomer structures, ground-truth dimer complexes, predicted structures, and apo conformations—the first dataset to pair predicted and apo structures for training flexible docking methods. The dataset is approximately 500 times larger than previous state-of-the-art datasets.

The Python API handles automatic downloading and caching of dataset files to a local directory (defaulting to ~/.local/share/pinder), with command-line tools for managing downloads and updates. It depends on 19 runtime packages including torch, torch-geometric, biotite, and pandas, making it suitable for machine learning workflows. The dataset requires approximately 700 GB of disk space when fully unpacked.

Use it for:

• Train protein-protein docking models using the large paired dataset of holo and apo structures
• Benchmark docking algorithms against gold-standard test sets included in the dataset
• Access preprocessed protein structure data and metadata for structural biology research
• Develop flexible docking methods using predicted and experimental structure pairs
• Evaluate protein interaction prediction models on standardized benchmarks

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Pinder provides access to a large protein-protein interaction dataset and tools for training and evaluating protein docking algorithms, including paired predicted and apo structures for flexible docking.

Yes, if you are actively developing or benchmarking protein docking algorithms and have approximately 700 GB of disk space available. The dataset is substantially larger than prior resources and uniquely includes paired predicted and apo structures. However, maintenance is dormant, so verify compatibility with your current PyTorch and torch-geometric versions before committing to a production workflow. For exploratory work or small-scale evaluation, the download overhead may not justify the install.

Install

pinder on PyPI

pip install pinder

uv add pinder

poetry add pinder

Installing pinder

Before you install

Low install friction with a pure Python wheel available. The package depends on fastpdb, which has pre-built wheels for Linux (glibc≥2.34), macOS Sierra or newer, and Windows; other platforms require building from source with the Rust toolchain. Maintenance is dormant—last release was 2024-11-15.

License in practice

Licensed under Apache 2.0 (permissive), allowing commercial and private use with minimal restrictions.

Quickstart

pip install pinder
from pinder.core import get_pinder_location
get_pinder_location()

Requires Python ≥3.10. The dataset itself is approximately 700 GB and must be downloaded separately via pinder_download command-line tool or manually from Google Cloud Storage. fastpdb may require Rust toolchain on unsupported platforms.

Verify before relying

• Whether dormant maintenance status affects stability or compatibility with current PyTorch/torch-geometric versions
• Performance characteristics and typical memory footprint when loading subsets of the dataset
• Whether optional dependencies (pytorch-cluster, PRODIGY-cryst) are commonly needed for typical workflows

Package facts

Evidence: pinder-0.5.0-py3-none-any.whl

Tags

More Artificial Intelligence packages