Unveiling the power of high-dimensional cytometry data with cyCONDOR

Abstract

High-dimensional cytometry (HDC) is a powerful technology for studying single-cell phenotypes in complex biological systems. Although technological developments and affordability have made HDC broadly available in recent years, technological advances were not coupled with an adequate development of analytical methods that can take full advantage of the complex data generated. While several analytical platforms and bioinformatics tools have become available for the analysis of HDC data, these are either web-hosted with limited scalability or designed for expert computational biologists, making their use unapproachable for wet lab scientists. Additionally, end-to-end HDC data analysis is further hampered due to missing unified analytical ecosystems, requiring researchers to navigate multiple platforms and software packages to complete the analysis. To bridge this data analysis gap in HDC we develop cyCONDOR, an easy-to-use computational framework covering not only all essential steps of cytometry data analysis but also including an array of downstream functions and tools to expand the biological interpretation of the data. The comprehensive suite of features of cyCONDOR, including guided pre-processing, clustering, dimensionality reduction, and machine learning algorithms, facilitates the seamless integration of cyCONDOR into clinically relevant settings, where scalability and disease classification are paramount for the widespread adoption of HDC in clinical practice. Additionally, the advanced analytical features of cyCONDOR, such as pseudotime analysis and batch integration, provide researchers with the tools to extract deeper insights from their data. We use cyCONDOR on a variety of data from different tissues and technologies demonstrating its versatility to assist the analysis of high-dimensional data from preprocessing to biological interpretation.

Fig. 1. Overview of the cyCONDOR ecosystem.

a The cyCONDOR ecosystem accepts HDC data from a variety of technologies combined with sample annotation. b The ecosystem covers a broad variety of analytical tasks, from data import and transformation to ML-based sample classifiers. Created in BioRender. Bonaguro, L. (2024) https://biorender.com/h88w007.

Fig. 2. cyCONDOR workflow for data pre-processing and annotation.

a Schematic overview of the first steps of cyCONDOR analysis, from data ingestion to cell labeling. b Pseudobulk Principal Component Analysis (PCA) colored by experimental groups. c Heatmap showing mean marker expression for each samples, column order is defined by hierarchical clustering. d UMAP colored by experimental group. e UMAP colored according to Phenograph clustering. f UMAP colored according to cell type annotation and heatmap of mean marker expression for each cell type, color coding legend is shared for both. g UMAP visualization of SpectralFlow data colored by Phenograph clustering. h UMAP visualization of CyTOF data colored by Phenograph clustering. i UMAP visualization of CITE-seq data colored by Phenograph clustering. Created in BioRender. Bonaguro, L. (2024) https://biorender.com/o70s217.

Fig. 3. Technical differences between batches can be mitigated with cyCONDOR.

a Schematic overview of the batch correction workflow implemented in cyCONDOR. b Original UMAP colored according to the experimental batch (left) and split by the experimental batch (right). c Batch corrected UMAP colored according to the experimental batch (left) and split by the experimental batch (right). d Original UMAP colored by Phenograph clustering. e Batch corrected UMAP colored by Phenograph clustering. Created in BioRender. Bonaguro, L. (2024) https://biorender.com/l04e722.

Fig. 4. Pseudotime inference on cytometry data helps to describe continuous developmental processes.

a Schematic overview of the subsetting workflow implemented in cyCONDOR. b UMAP of all bone marrow cells colored by annotated cell type. c UMAP of the subset of monocytes, pDCs and their progenitors colored by annotated cell type. d Schematic overview of the pseudotime inference workflow implemented in cyCONDOR. e UMAP colored according to the inferred pseudotime of the predicted trajectories. f Heatmap of protein expression in cells belonging to the monocytes trajectory ordered according to the inferred pseudotime. Created in BioRender. Bonaguro, L. (2024) https://biorender.com/x99o393.

Fig. 5. cyCONDOR provides accessible function for differential analysis.

a Schematic overview of the differential analysis workflow. b UMAP of the PBMCs dataset colored by annotated cell type,s color coding shared with 5c. c Confusion matrix of the annotated cell types split by experimental group. d Stacked barplot of the cellular frequencies of the annotated cell types split by experimental groups. e Boxplot of the frequency of each annotated cell type split by experimental group (Ctrl, n = 3; Dis. n = 3, n number defines individual donors, Tukey-style boxplot). f Heatmap of the average expression of each marker split by cell type and experimental group. Statistical significance was calculated with a two-sided t-test with default settings and bonferroni multiple test correction, *p < 0.05, **p < 0.01, ***p < 0.001 (Unconventional T cells p = 0.04917). Source data are provided as a Source Data file. Created in BioRender. Bonaguro, L. (2024) https://biorender.com/v89t374.

Fig. 6. Batch alignment allows accurate analysis of longitudinal data.

a Schematic overview of the data projection workflow implemented in cyCONDOR. b UMAP visualization of the training dataset colored according to the annotated cell type. c UMAP overlapping the projected data (purple) to the training dataset (gray). d LISI scores calculated between training and projected data. e Left: UMAP visualization of the projected data colored according to the predicted cell types, right: UMAP of the original data colored by cell label used to train UMAP model and kNN classifier. f confusion matrix comparing the manual annotation of the projected data with the predicted cell labels. Created in BioRender. Bonaguro, L. (2024) https://biorender.com/a35t149.

Fig. 7. Direct implementation of clinical classifier allows the accurate classification of disease states.

a Schematic overview of the clinical classifier workflow implemented in cyCONDOR. b UMAP visualization of the training dataset colored by experimental groups. c Single-cell level probability for the test dataset split by sample and colored by experimental group. d Sample level probability for the test dataset split by sample and colored by experimental group. e Accuracy, specificity and sensitivity of a clinical classifier trained on the entire FlowCap-II dataset (100 permutations, Tukey-style boxplot). Source data are provided as a Source Data file. Created in BioRender. Bonaguro, L. (2024) https://biorender.com/u02j561.