Studies on cell death types, especially in cancer and degenerative diseases, hold promise for disease therapies. However, identifying specific cell death types is challenging, requiring intensive computational work and costly assays. To address this,  Kenji Schorpp, Alaa Bessadok, Aidin Biibosunov, Ina Rothenaigner, Stefanie Strasser, Tingying Peng, and Kaymar Hadian from the Research Unit Signaling and Translation, Molecular Targets and Therapeutics Center and Helmholtz AI at Helmholtz Munich have created CellDeathPred. 

CellDeathPred is a deep-learning framework using high-content imaging to distinguish ferroptotic and apoptotic cells from healthy ones. It incorporates a deep neural network to classify learned images into cell death modalities and optimizes learning with the supervised contrastive loss function. CellDeathPred's efficacy was demonstrated separating cell groups with 95% accuracy, supporting its capacity for cell death discovery. 

The proposed framework learns directly from cell painting images, eliminating the need for pre-filtering. This simplifies computation and allows the model to handle diverse cell images, making it more versatile for different cell death-related images. The study aims to answer whether high-content cell painting assay images can classify drug-induced ferroptosis, apoptosis, or no adverse effect. This prediction could be valuable in identifying cell death types in clinically relevant drugs, leading to potential new therapeutic opportunities.

This work has been published in Cell Death Discovery.

Representative image of the cell painting for an Apoptosis (left) and Ferroptosis (right) inducers. (Schorpp, K., Bessadok, A., Biibosunov, A. et al. 2023)

Original publication

Schorpp, K., Bessadok, A., Biibosunov, A. et al. CellDeathPred: a deep learning framework for ferroptosis and apoptosis prediction based on cell painting. Cell Death Discov. 9, 277 (2023). https://doi.org/10.1038/s41420-023-01559-y