New open-source platform supports large imaging data analysis

Modern day drug discovery is shifting from single end point assays to more complex phenotypic assays that query single cell and population responses to chemicals and genetic manipulation. One such assay, cell painting, is designed to use imaging to highlight cellular substructures and, via image analysis pipelines, to quantify changes in the cellular state. This type of analysis often requires powerful computational resources and results in very large datasets that are difficult to interpret easily at the individual cell level, resulting in data averaging that can obscure the underlying heterogeneity of cell population responses.

A new open-source image analysis platform, developed by researchers at Baylor College of Medicine, Texas A&M University and the University of Houston, will now provide researchers with a powerful tool to analyze these large datasets in a computational resource efficient way, while including evaluation of diverse single cell responses among heterogeneous populations.  

The platform, called SPACe (Swift Phenotypic Analysis of Cells), addresses a significant limitation facing academic labs and small institutions: the computational resources required to analyze large volumes of imaging data. This new platform is described in the latest edition of Nature Communications.    

“The pharmaceutical industry has been accustomed to simplifying complex data into single metrics. This platform allows us to shift away from that approach and instead capture the full diversity of cellular responses, providing richer, more informative data that can reveal new avenues for drug development,” said Dr. Michael Mancini, professor of molecular and cellular biology and director of the Gulf Coast Consortium Center for Advanced Microscopy and Image Informatics co-located at Baylor College of Medicine and TAMU Institute for Bioscience and Technology. “This new platform is open-source and available to anyone. We see this impacting both academic and pharmaceutical research communities.”

While pharmaceutical companies have historically relied on high-powered cloud computing systems to analyze this data, the new platform is designed to be accessible even to researchers using standard computers, lowering the barriers to entry for sophisticated cellular analysis.

At its core, the platform improves on existing methods by enabling the analysis of thousands of individual cells generated by increasingly faster automated imaging platforms, and better capturing the variability of biological processes. This innovation enables a more nuanced understanding of how drugs interact with cells, revealing insights into mechanisms beyond cell death, such as changes in cell and individual organelles (nucleus, nucleolus, mitochondria, cytoskeleton, cytoplasm) phenotype. Collectively, this additional information adds an important, expanded dimension that can facilitate an increased understanding of a drug’s mechanism of action.

“The platform allows for the identification of non-toxic effects of drugs, such as alterations in cell shape or effects on specific organelles, which are often overlooked by traditional assays that focus largely on cell viability,” said Dr. Fabio Stossi, currently a senior scientist with St. Jude Children’s Research Hospital, the lead author formerly with Baylor during the development of this platform.

He added that the SPACe can analyze thousands of cells, allowing for large-scale drug screenings on a standard computer, which makes this process available to laboratories of varying sizes and helps more researchers work together.

"This tool could be a game-changer in how we understand cellular biology and discover new drugs,” Stossi said. “By capturing the full complexity of cellular responses, we are opening new doors for drug discovery that go beyond toxicity.”

“The platform incorporates state-of-the-art routines for cell detection and feature extraction that was implemented in Python, ensuring high computational efficiency, portability and additional flexibility,” said Demetrio Labate of the University of Houston.

Researchers interested in using the platform can access it through Github at https://github.com/dlabate/SPACe. The team plans to continue expanding its capabilities through collaborations with other institutions and research centers.

For more information, including access to the platform and the published paper, please visit 
https://www.nature.com/articles/s41467-024-54264-4.  

Others who contributed to the research and development of SPACe include: Pankaj K. Singh, Michela Marini, Kazem Safari, Adam T. Szafran, Alejandra Rivera Tostado, Christopher D. Candler, Maureen G. Mancini, Elina A. Mosa, Michael J. Bolt and Demetrio Labate. They are affiliated with Baylor College of Medicine, Texas A&M University or University of Houston.

Software development, experimental approaches and imaging for this project were supported by the GCC Center for Advanced Microscopy and Image Informatics (CAMII, CPRIT RP170719) and the Integrated Microscopy Core at Baylor College of Medicine (funding from NIH (DK56338, CA125123, ES030285, 699 S10OD030414) and CPRIT (RR200043).