AB1334 AN OPEN-SOURCE WEB-BASED ANALYTICS PLATFORM FOR THE HUMAN IMMUNE ATLAS

Abstract

Background: Mass cytometry (CyTOF) measures the expression of many proteins (currently up to 40) in single cells. TII applies CyTOF to characterise cellular diversity of the immune system in peripheral blood and other tissues. Using a growing collection of CyTOF data acquired from blood of clinically and demographically diverse human subjects, we have built our version of the immune atlas, called EPIC (Extended Poly-dimensional Immunome characterisation). Objectives: We will discuss two main objectives of this project. To integrate CyTOF data, metadata, cell annotations or other inferred data into the immune atlas, we developed a novel data analytics pipeline. To provide a user-friendly gateway that helps researchers explore the immune atlas and gain new insights about their own CyTOF data, we implemented a web-based data analytics application using the R Shiny programming environment. Methods: The core components of the atlas are ‘immune maps’, which comprise CyTOF data of samples labeled with identical antibody panels and grouped according to a common biological theme. Besides protein expression patterns, immune maps contain contain clinical and demographic metadata, as well as phenotypic information inferred from clustering and cell type annotation. To intuitively analyse these complex multi-dimensional data, we developed a Shiny/R web application that has two main objectives. First, clients can explore the immunome at different levels of details using a wide range of interactive visualisation methods, such as bar charts comparing the abundance of all or subsets of immune cell types in different age groups, or tSNE/UMAP scatter plots providing global perspectives of expression domains. Second, users can upload their own CyTOF data and, through pattern matching, obtain instant estimates about the abundance of selected immune cell populations in their own samples. Results: We tested our system using immune maps constructed from healthy paediatric samples. Manually gated ground truth data along with interactive visualisation techniques were used to measure the accuracy of our pipeline in detecting and annotating homogenous cell populations. In addition, we will demonstrate how immune maps can be applied to classify uploaded CyTOF data. Conclusion: Our interactive immune atlas platform promises to improve our understanding of the changing immunome landscape in response to disease, treatment and ageing. References NA Disclosure of Interests: Martin Wasser: None declared, Joo Guan Yeo: None declared, Pavanish Kumar: None declared, Thaschawee Arkachaisri Speakers bureau: Abbvie Pte, Ltd, Su Li Poh: None declared, Fauziah Ally: None declared, Jing yao Leong: None declared, Kee Thai Yeo: None declared, Salvatore Albani: None declared