test_output_README.md

Test Output Directory

This directory contains persistent test output files generated during test runs. Unlike temporary directories, these files are preserved after tests complete to allow manual inspection and debugging.

Directory Structure

test_output/
├── test_<timestamp>/          # Individual test run directories
│   ├── *.npz                  # AlphaGenome prediction data (NumPy archives)
│   ├── *.json                 # Metadata and configuration files
│   ├── *.h5ad                 # AnnData files for variant scoring
│   └── *_files.json          # File lists for variant predictions
└── prediction_result_*.npz    # Model test files
└── variant_result_*.json      # Model test files
└── scoring_result_*.json      # Model test files

File Types

• .npz files: Compressed NumPy archives containing genomic track data (values, metadata, intervals)
• .json files: Human-readable metadata, configuration, and file lists
• .h5ad files: AnnData format for variant scoring results
• *_files.json: File lists linking reference/alternate predictions

Cleanup

The directory is automatically cleared at the start of each test session but preserved after completion. To manually clean up:

rm -rf test_output/

Inspection

You can examine the files to understand:

• What data AlphaGenome predictions generate
• File formats and structures used
• Metadata included with predictions
• File organization patterns

Generated By

These files are created by the test suite in tests/test_server.py and represent realistic AlphaGenome MCP server outputs for debugging and validation purposes.