CyTOFmerge: integrating mass cytometry data across multiple panels

Abstract

Motivation: High-dimensional mass cytometry (CyTOF) allows the simultaneous measurement of multiple cellular markers at single-cell level, providing a comprehensive view of cell compositions. However, the power of CyTOF to explore the full heterogeneity of a biological sample at the single-cell level is currently limited by the number of markers measured simultaneously on a single panel.

Results: To extend the number of markers per cell, we propose an in silico method to integrate CyTOF datasets measured using multiple panels that share a set of markers. Additionally, we present an approach to select the most informative markers from an existing CyTOF dataset to be used as a shared marker set between panels. We demonstrate the feasibility of our methods by evaluating the quality of clustering and neighborhood preservation of the integrated dataset, on two public CyTOF datasets. We illustrate that by computationally extending the number of markers we can further untangle the heterogeneity of mass cytometry data, including rare cell-population detection.

Availability and implementation: Implementation is available on GitHub (https://github.com/tabdelaal/CyTOFmerge).

Supplementary information: Supplementary data are available at Bioinformatics online.

Fig. 1.

CyTOFmerge pipeline: split the sample, stain each partial sample with a different marker panel and apply CyTOF to obtain the panels’ measurements. Both panels A and B share a set of markers m (green). L1 (red) are unique markers of panel A, and L2 (blue) are unique markers of panel B. Both panel measurements are combined to obtain an extended markers measurements per cell, which is input to downstream computational analysis as, e.g. clustering in a t-SNE mapped domain shown here

Fig. 2.

Shared markers for the HMIS dataset. The selected markers that can best represent the dataset using (A) PCA, (B) AE and (C) HSNE (marker ordering is based on the PCA selection profile, black is selected, white is not selected)

Fig. 3.

Clustering of the original and the imputed datasets. (A–C) t-SNE maps showing the different identified populations in the CD4+T Cells lineage. (A) Shows the populations of the original data. (B) The populations of the imputed data (for m = 16, L1 = 6 and L2 = 6). (C) The mapping of the original clusters labels on the t-SNE map of the imputed data. (D) Heatmap of markers expression for the 121 characterized immune cells populations of the original dataset for m = 16. Black-to-yellow scale shows the median arcsinh-5 transformed values for the markers expression. Markers colors indicate whether a marker is shared between panels or unique to a single panel, during panels combination (red is shared, green is unique to panel A, blue is unique to panel B)

Fig. 4.

Marker panel extension impact on the identification of distinct populations in the TCRγδ immune lineage—panel A. (A) The Reduced t-SNE map using only 22 markers. (B) The original t-SNE map using the original 28 markers. (C) The imputed t-SNE map using 28 markers of which 6 are imputed from panel B. All three maps are colored with the original population labels. (D) Shared and missing markers expression profiles are shown on the original t-SNE map. The map border color indicate whether a marker is shared between panels or unique to a single panel (red is shared, green is unique to panel A, blue is unique to panel B and thus missing markers for panel A).The color bar shows the arcsinh-5 transformed values for the markers expression

Fig. 5.

Marker panel extension impact on the identification of distinct populations in the TCRγδ immune lineage—panel B. (A) The Reduced t-SNE map using only 22 markers. (B) The original t-SNE map using the original 28 markers values. (C) The imputed t-SNE map using 28 markers of which 6 are imputed from panel A. All three maps are colored with the original populations labels. (D) Shared and missing markers expression profiles are shown on the original t-SNE map. The map border color indicate whether a marker is shared between panels or unique to a single panel (red is shared, green is unique to panel A and thus missing markers for panel B, blue is unique to panel B).The color bar shows the arcsinh-5 transformed values for the markers expression