SAbDab CLI

A fast, multithreaded CLI for downloading data from SAbDab (the Structural Antibody Database). This tool is an alternative to the SAbDab downloader script.

Here is sabdab-cli in action:

Highlights

• ⚡ Concurrent Downloads: High-performance async I/O for rapid data acquisition.
• 🛡️ Robust Retries: Built-in exponential backoff for transient network issues.
• 📊 Rich Progress: Beautiful terminal progress bars and informative feedback.
• 📦 Atomic Writes: Prevents data corruption by using temporary files during downloads.

Installation

With uv (recommended):

uv add sabdab-cli

With pip:

pip install sabdab-cli

Or run directly without installing:

uvx sabdab-cli download -s summary.csv -o ./data --original-pdb

HTTP/2 Support (Optional)

For potentially faster downloads using HTTP/2, install with the http2 extra:

uv add "sabdab-cli[http2]"

Usage

You can invoke with sabdab-cli or sabdabc.

Download selected data types for entries in a SAbDab summary file:

sabdab-cli download \
  --summary-file summary.csv \
  --output-path ./data \
  --original-pdb \
  --chothia-pdb \
  --sequences \
  --annotation \
  --threads 10

Available Data Types

Summary File Format

The --summary-file (or -s) argument should be a tab-separated or comma-separated file containing at least the following columns: pdb, Hchain, Lchain, and model. The CLI automatically detects the delimiter based on the file content.

See the test summary file for an example.

Performance Tuning

By default, sabdab-cli uses adaptive concurrency based on your CPU count. You can manually adjust this for your environment:

# High-concurrency mode (e.g., 20 threads)
sabdabc download -s summary.csv -o ./data --original-pdb --threads 20

# Synchronous mode (single thread)
sabdabc download -s summary.csv -o ./data --original-pdb --threads 1

# Enable HTTP/2 for faster downloads (requires sabdab-cli[http2])
sabdabc download -s summary.csv -o ./data --original-pdb --http2

License

MIT