From Cells to Morphological Profiles A FAIR cloud processing pipeline using Galaxy

From Cells to Morphological Profiles A FAIR cloud processing pipeline using Galaxy Arsenio N. Cruz C.1, Thanh Huy Pham1, Collin Zimmer1, Riccardo Massei2, Yi Sun3, Paul Czodrowski1 1 Chemistry Department, Johannes Gutenberg University, Mainz, Germany 2 Department of Monitoring and Exploration Technologies, Helmholtz-Zentrum für Umweltforschung - UFZ, Leipzig, Germany 3 Data Science Centre, European Molecular Biology Laboratory (EMBL), Heidelberg, Germany Profiles postprocessing[5–7] I) Feature aggregation Aggregate single-cell features at well-level. III) Linear transformation Use PCA/UMAP to cluster features for visualization. II) Feature reduction • Retain reproducible features. • Keep biologically relevant features. • Avoid correlated features. Feature n [Control] Feature n [Compound m] EMD Z-score Aggregated +2.3 0.8 Ensemble measurements • Z-scores • Earth Mover’s Distance (EMD) Statistical metrics for distributions Capture non-Gaussian distributions. Suitable for Gaussian distributions. MoA with good matching will cluster. Wells with good replicate will cluster. Profile similarity metrics Percent matching Proportion of compounds matching replicates of itself. Percent replicating Proportion of compounds matching those with the same MoA. II) Object identification III) Feature measurement Nuclei CytoplasmCell area Cytoplasmic area Cytoplasmic perimeter Nuclear area Nuclear perimeter Nuclear aspect ratio (width/length) Number of mitochondria Number of vesicles per cell ... [227.6, 58.6, 73.2, 24.6, 1.3, 7, 4, ...] Feature extraction[3][4] I) Image preprocessing Illumination correction, quality control • High-throughput phenotypic screening to gain insights about an analyte‘s cellular mode of action (MoA)[3] • Chemical or biological perturbations can be investigated. We mainly employ libraries of natural products with yet uncharacterized MoA. • Cells are fixed, permeabilized, and stained (Hoechst 33342, ConA-AF488, Syto14, Phalloidin-AF568, WGA-AF555, MitoTracker) • Applicable to robust cell models or disease-relevant primary cells • Suitable replicate layout, illumination correction, and reference analytes enable quality control to ensure reliable data nucleus (DNA) nucleoli, rough ER (RNA) endoplasmatic reticulum mitochondria cytoskeleton, plasma membrane and Golgi microscopy channels visualized compartments perturbation cell model morphological changes 9 fields of view 3 stacks of view cppipe file CellProfiler headless and interactive tool External image raw data storage Data Processing Workflow Results Galaxy is an online computational platform used by a global community of thousands of scientists for processing of large-scale data. This collective effort includes the development of the Galaxy software framework, the integration of analysis tools and visualizations, and the operation of public servers that provide access to Galaxy through web browsers (usegalaxy.eu) Galaxy increases the FAIRness of data analysis pipelines by providing versioned tools and workflows that can be annotated, shared and published. Furthermore, Galaxy has a dedicated interface for image data analysis—imaging.usegalaxy.eu—providing a comprehensive suite of tools and workflows tailored specifically for imaging scientists. Through the implementation of open source CellProfiler tool into the Galaxy framework, we are creating an interactive evaluation environment for the Cell Painting community.[1][2] This environment can aid the drug discovery process. Overview on the Cellpainting processing pipeline using Galaxy. Images and Cellprofiler pipeline are uploaded to usegalaxy.eu and processed using the CellProfiler headless tool. Workflow can be then uploaded to the WorkflowHub for automatic execution. [1]Massei, R.et al. (2025). Development FAIRimage analysis workflows and RDM pipelines in Galaxy. 2nd Conference on ResearchData Infrastructure (CoRDI),Aachen, Germany. [2]Hiltemann, Saskia,Rasche,Helena et al. (2023) Galaxy Training: A Powerful Framework for Teaching! PLOSComputational Biology [3]Bray, M. A., et al. (2016) Cell Painting, a high-content image-based assayfor morphological profiling using multiplexed fluorescent dyes. Nature Protocols, 11, 1757–1774 [4]Stirling, D. R., et al. (2021). CellProfiler 4: improvements in speed, utility and usability. BMC Bioinformatics, 22(1), 433 [5]Caicedo, J. C.,et. al. (2017). Data-analysis strategies for image-based cell profiling. Nature Methods, 14(9), 849–863. [6]Pearson,Y.E.,et. al. (2022).A statistical framework for high-content phenotypic profiling using cellular feature distributions. Communications Biology, 5(1). [7]Cimini, B.A., et. al. (2023). Optimizing the Cell Painting assayfor image-based profiling [Supplementary information]. Nature Protocols, 18(7), 1981–2013. Images obtained for this poster are under CC0license (image source: www.svgrepo.com). The software Cellprofiler is distributed under BSDlicense. NFDI4BIOIMAGEisfunded by the Deutsche Forschungsgemeinsschaft (DFG,German ResearchFoundation) under the National ResearchDataInfrastructure, grant number NFDI46/1, project number 501864659