CellDeathPred: a deep learning framework for ferroptosis and apoptosis prediction based on cell painting

Area of Science:

• Biomedical imaging
• Computational biology
• Cellular biology

Background:

• Cell death, including apoptosis and ferroptosis, is crucial in development, homeostasis, and disease pathogenesis.
• Identifying specific cell death types is complex, requiring computationally intensive methods and expensive experiments.
• Targeting cell death pathways offers therapeutic potential for diseases like cancer and neurodegeneration.

Purpose of the Study:

• To develop a novel deep-learning framework, CellDeathPred, for distinguishing between ferroptosis, apoptosis, and healthy cells.
• To leverage high-content imaging and cell painting techniques for automated cell death classification.
• To provide a computationally efficient and accurate method for cell death discovery.

Main Methods:

• Utilized a deep neural network to embed microscopic cell painting images into a discriminative latent space.
• Implemented a supervised contrastive loss function for optimizing the deep learning model.
• Trained and validated the CellDeathPred framework on microscopy datasets of human HT-1080 cells induced into ferroptosis and apoptosis.

Main Results:

• CellDeathPred successfully distinguished ferroptotic and apoptotic cells from healthy controls.
• Achieved an average accuracy of 95% on non-confocal cell painting datasets.
• Demonstrated the framework's capability in classifying different cell death modalities.

Conclusions:

• CellDeathPred offers a robust and accurate deep-learning approach for cell death classification using cell painting.
• The framework addresses the challenge of identifying cell death types, facilitating disease research and therapeutic development.
• Supports the potential of AI-driven image analysis in advancing cell death discovery and understanding disease mechanisms.