Abstract LB252: iCellSight: An interactive tool for analyzing and visualizing in-situ high-plex cellular protein data

Abstract

Abstract Introduction The increased use of tissue-based multiplex immunohistochemistry (mIHC) in cancer research has significantly enhanced the generation of spatial hyper-plex images, crucial for understanding complex tumor-immune microenvironments. This advancement has led to a growing demand for spatial data analysis tools that enable quick and insightful evaluation. While existing spatial analysis software like Steinbock and SPIAT offer foundational support, effectively managing the high-dimensional and voluminous cellular data to derive meaningful biological insights remains challenging. This is especially true for researchers lacking bioinformatics expertise who seek to conduct quick preliminary evaluations. To address this gap, we introduce iCellSight: a user-friendly, code-free interactive analysis and visualization tool tailored to empower researchers with rapid and intuitive interpretation of in-situ cellular data. Material and methods A formalin-fixed, paraffin-embedded colorectal carcinoma (CRC) patient tissue section was subjected to staining using COMET (Lunaphore, Switzerland), which automates sequential immunofluorescence staining and imaging of up to 40 protein markers. The stacked image file produced was analyzed with HALO (Indica Labs, USA), which generated cell-level data like cell coordinates, stain intensity values and user-defined cell types. We developed a Shiny App tool, using the Shiny R package, with key features for cell data interpretation: (1) user-input cell number subsampling; (2) dimensional reduction plots with PhenoGraph for unsupervised clustering, allowing a choice of UMAP or t-SNE, and neighbor number selection of 30, 50 or 150; (3) 2D tissue space mapping of PhenoGraph clusters; and (4) heatmap visualizations displaying cluster-wise biomarker expression (with customizable color schemes and scaling), biomarker co-expression, and cluster pair-wise distance. Results Using iCellSight, 297,000 cells were detected from the CRC sample, where 27 clusters were identified. We observed 3 distinct clusters from the t-SNE plot that consist of macrophages, granulocytes, and NK cells and monocytes respectively. From the cluster projection in 2D tissue space, the cluster of NK cells and monocytes was found to be densely populated in a particular area of the tumor nest. Conclusion We presented iCellSight, a code-free visualization and analysis tool which streamlines the semi-automated analysis of spatial hyper-plex protein data. Designed with an intuitive user interface, this app enables researchers to efficiently navigate its features without requiring any prior coding expertise. We anticipate that the visualizations and analysis provided by this tool will significantly assist researchers in their data analyses and facilitate the validation of experimental hypotheses. Citation Format: Yancun Zhu, Felicia Wee, Menaka Priyadharsani Rajapakse, Esther Gek Teo, Nicholas Ang, Solomonraj Wilson, Zheng Yi Ho, Willa Wen-You Yim, Li Yen Chong, Craig Ryan Joseph, Jeffrey Chun Lim, Zhen Wei Neo, Chwee Ming Lim, Bernett Lee, Olaf Rotzschke, Joe Yeong, Mai Chan Lau. iCellSight: An interactive tool for analyzing and visualizing in-situ high-plex cellular protein data [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(7_Suppl):Abstract nr LB252.