An Automated Cell Line Authentication Method for AstraZeneca Global Cell Bank

This is the official public Pytorch implementation for our paper (https://www.nature.com/articles/s41598-022-12099-3).

For any issue and question, please email [[email protected]]

Dependencies

• Python (>=3.6)
• Pytorch (>=1.9.0)
• opencv-python
• matplotlib
• scikit-learn (>=0.24.2)
• numpy
• scikit-image
• torchvision (>=0.10.0)

Dataset

Part of example images are put in the './data'.

Training CLCNet

cd ./networks/classification
python cell_classification.py --bs 20 --arch Xception 

Testing CLCNet

Use JupyterLab to open 'model_evaluation.ipynb' and run code blocks.

Training/Testing CLRNet or with tranfer learning

Use JupyterLab to open 'model_evaluation.ipynb' and run all code blocks.

License

This code is made available under the GPLv3 License and is available for non-commercial academic purposes.

Acknowledgement

The authors gratefully acknowledge financial support from University of Leicester, AstraZeneca UK, China Scholarship Council.

---

If you find that is useful in your research, please consider citing:

@article{tong2022automated,
  title={An automated cell line authentication method for AstraZeneca global cell bank using deep neural networks on brightfield images},
  author={Tong, Lei and Corrigan, Adam and Kumar, Navin Rathna and Hallbrook, Kerry and Orme, Jonathan and Wang, Yinhai and Zhou, Huiyu},
  journal={Scientific Reports},
  volume={12},
  number={1},
  pages={1--11},
  year={2022},
  publisher={Nature Publishing Group}
}