Quick start
===========

This example reuses the supplied three-channel masks, builds meshes, curates
the cells, detects GFP-DnaN objects, and saves morphology and signal measurements.
The script first copies the dataset so that the original inputs and outputs
remain available for comparison.

Run this in the BactoScoop environment, or in its Jupyter kernel:

.. literalinclude:: ../_downloads/quickstart.py
   :language: python

Download :download:`the complete script <../_downloads/quickstart.py>`.
The default calibration matches the supplied walkthrough; replace it with your
microscope's µm/pixel calibration for your own images.

What to inspect
---------------

Inspect the mask outlines before trusting the reconstructed meshes. After
curation, compare the kept and rejected contours. Then inspect individual
object detections and check whether the merged table contains the expected
channels. See :doc:`../examples/plotting` for the plotting functions.

This quick start intentionally uses the supplied SVM with its matching dataset.
The same model need not classify cells correctly in a different experiment.
Review :doc:`../workflow/curation` before applying it to other images.

Expected files
--------------

The working copy contains new ``_meshdata.pkl``, ``_curated_meshdata.pkl``,
``_features.pkl``, and ``_features.parquet`` files. The script also saves
``processing_summary.json`` so item-level and cell-level failures can be reviewed.
Existing masks remain in the working copy's ``masks/`` folder.