High throughput single-cell analysis pipeline with ML-powered adaptive k value thresholding for spot counting of FISH spots from Amnis true imaging flow cytometers
INTRODUCTION What is the FISHSpotCounter project?
Running the application: Where to start?
TO BE ADDED IN THE FUTURE
File layout:
project_root/
│
├── core/
│ ├── image_processing.py # Filtering, thresholding, pre/post-processing
│ ├── spot_counter.py # detect_spots logic
│ └── file_handler.py # File loaders, normalisation, etc.
│
├── estimator/
│ └── k_predictor.py # Load model and batch-predict K values
│
├── optimiser/
│ └── k_optimiser.py # Recursive/bimodal-aware optimiser for model training
│
├── training/
│ ├── feature_extraction.py # Converts image + K into features
│ └── model_training.py # Trains ML models such as XGBoost (default) or similar
│
├── gui/
│ └── SpotCounterGUI.py # GUI frontend using tkinter, PyQt, etc. for the app
│
├── main.py # Entrypoint with argument parsing
├── config.py
└── requirements.txt # Dependencies list, PIP install -r compatible
Other files: README.md --> opens this readme file! DevelopmentNotes.md --> a record and notes during the development of this app .gitattributes --> hidden file for git version control