dragonfly-automation

This is a Python package that contains the microscope-automation scripts used to enable high-throughput microscopy for the OpenCell project. These scripts use mm2python to control the spinning disk confocal microscope (nicknamed 'dragonfly'). They are written against the MicroManager 2.0 beta API.

Refer to our 2022 OpenCell paper for more details.

Primary features

• Modular and reusable methods for common tasks (e.g., autofocusing, adjusting exposures, acquiring z-stacks)
• Pre-trained machine-learning models for dynamic field-of-view scoring and/or classification
• Extensive logging of metadata, errors, and MicroManager API calls
• Mocks for the MicroManager APIs to facilitate development and enable testing

Usage examples

The script is called using a CLI. Some common examples are shown below. See scripts/run_acquisition.py for all CLI arguments.

// start an acquisition in test mode and using the mocked MicroManager API
// (this does not require mm2python or MicroManager)
python ./scripts/run_acquisition.py \
--data-dir ./tests/output/ \
--pml-id PML0000 \
--plate-id P0000 \
--platemap-type none \
--mode test \
--mock-micromanager-api

// start a full-plate acquisition, using mm2python and the real MicroManager API
python ./scripts/run_acquisition.py \
--pml-id PML0123 --plate-id P0021 --platemap-type none

Development

We use black for formatting, flake8 and pylint for linting, pytest for testing, and Github Actions for CI. To setup a new dev env:

1. Create a new Python 3.7 virtualenv: conda create -n dragonflyenv python=3.7
2. Clone this repo and install dependencies using pip: pip install -r requirements.txt
3. Install the dragonfly_automation package: pip install -e .
4. Install the pre-commit hooks: pre-commit install
5. To run tests locally, use make test