debase 0.5.0

Enzyme lineage analysis and sequence extraction package

DEBase

DEBase is a Python package for extracting and analyzing enzyme lineage data from scientific papers using AI-powered parsing.

Features

• Extract enzyme variant lineages from PDF documents
• Parse protein and DNA sequences with mutation annotations
• Extract reaction performance metrics (yield, TTN, ee)
• Extract and organize substrate scope data
• Match enzyme variants across different data sources using AI
• Generate structured CSV outputs for downstream analysis

Installation

pip install debase

Quick Start

# Run the complete pipeline
debase --manuscript paper.pdf --si supplementary.pdf --output results.csv

# Enable debug mode to save Gemini prompts and responses
debase --manuscript paper.pdf --si supplementary.pdf --output results.csv --debug-dir ./debug_output

# Individual components with debugging
python -m debase.enzyme_lineage_extractor --manuscript paper.pdf --output lineage.csv --debug-dir ./debug_output
python -m debase.reaction_info_extractor --manuscript paper.pdf --lineage-csv lineage.csv --output reactions.csv --debug-dir ./debug_output
python -m debase.substrate_scope_extractor --manuscript paper.pdf --lineage-csv lineage.csv --output substrate_scope.csv --debug-dir ./debug_output
python -m debase.lineage_format -r reactions.csv -s substrate_scope.csv -o final.csv -v

Debugging

Use the --debug-dir flag to save all Gemini API prompts and responses for debugging:

• Location extraction prompts
• Sequence extraction prompts (can be very large, up to 150K characters)
• Enzyme matching prompts
• All API responses with timestamps
• Note: lineage_format.py uses -v for verbose output instead of --debug-dir

Requirements

• Python 3.8+
• Google Gemini API key (set as GEMINI_API_KEY environment variable)

Version

0.4.4

License

MIT License

Authors

DEBase Team - Caltech

Contact

[e-mail]