Development

Development

A passing test can still encode the wrong LOD or LOQ. Coverage does not prove a figure of merit is defensible.

Install as on Installation. Then:

uv run pytest
uv run ruff check loqculate/ tests/
uv run ruff format --check loqculate/ tests/

CI runs those checks on Linux, macOS, and Windows.

Layout

CLI input -> io readers -> model fit -> LOD / LOQ -> CSV / plots

• loqculate/cli.py: user interaction. It calls readers and models.
• loqculate/config.py: defaults.
• loqculate/io/: readers, writers, multipliers, CalibrationData.
• loqculate/models/: fit / predict / lod / loq.
• loqculate/utils/: shared math.
• loqculate/compat/: paper-reproduction ports of old/.
• loqculate/plotting/: figures.
• loqculate/testing/: synthetic generators for tests and benches.

Documentation

Wiki pages are Markdown files in wiki/ in this repository.

Pull requests

Open pull requests against dev. main is the last release.