CytoNorm: A Normalization Algorithm for Cytometry Data

Abstract

High‐dimensional flow cytometry has matured to a level that enables deep phenotyping of cellular systems at a clinical scale. The resulting high‐content data sets allow characterizing the human immune system at unprecedented single cell resolution. However, the results are highly dependent on sample preparation and measurements might drift over time. While various controls exist for assessment and improvement of data quality in a single sample, the challenges of cross‐sample normalization attempts have been limited to aligning marker distributions across subjects. These approaches, inspired by bulk genomics and proteomics assays, ignore the single‐cell nature of the data and risk the removal of biologically relevant signals. This work proposes CytoNorm, a normalization algorithm to ensure internal consistency between clinical samples based on shared controls across various study batches. Data from the shared controls is used to learn the appropriate transformations for each batch (e.g., each analysis day). Importantly, some sources of technical variation are strongly influenced by the amount of protein expressed on specific cell types, requiring several population‐specific transformations to normalize cells from a heterogeneous sample. To address this, our approach first identifies the overall cellular distribution using a clustering step, and calculates subset‐specific transformations on the control samples by computing their quantile distributions and aligning them with splines. These transformations are then applied to all other clinical samples in the batch to remove the batch‐specific variations. We evaluated the algorithm on a customized data set with two shared controls across batches. One control sample was used for calculation of the normalization transformations and the second control was used as a blinded test set and evaluated with Earth Mover's distance. Additional results are provided using two real‐world clinical data sets. Overall, our method compared favorably to standard normalization procedures. The algorithm is implemented in the R package “CytoNorm” and available via the following link: http://www.github.com/saeyslab/CytoNorm © 2019 The Authors. Cytometry Part A published by Wiley Periodicals, Inc. on behalf of International Society for Advancement of Cytometry.

CytofRUV: Removing unwanted variation to integrate multiple CyTOF datasets

“…Assuming that all 20 clusters have cells from CLL1 and HC1 in both batches, any differences in protein expression between cells within the cluster but in different batches will represent unwanted variation. We summarized the performance of our CytofRUV normalisation method on all the CLL samples using three metrics and compare them with the corresponding ones for BatchAdjust (Schuyler et al, 2019) and CytoNorm (Van Gassen et al, 2019).…”

“…4). To quantify the batch differences between these pairs of distributions, we computed the Earth Movers Distance (EMD) as described in (Van Gassen et al, 2019) for all the proteins and all CLL and HC samples for both the original dataset and the normalized datasets ( Fig. 5A and 5D).…”

“…To expand our analysis, we also tested all methods on two other datasets containing replicate reference samples across multiple batches, taken from those used in (Schuyler et al, 2019) and in CytoNorm (Van Gassen et al, 2019).…”

“…However technical variation can impact specific cell types differently (Van Gassen, Gaudilliere, Angst, Saeys, & Aghaeepour, 2019). To address this, CytoNorm (Van Gassen et al, 2019) uses the clustering algorithm FlowSOM (Van Gassen et al, 2015) to identify cell subpopulations prior to normalisation, and defines a subpopulation-specific goal distribution for the values of each protein measurement using the means of quantiles. Their approach then uses splines to transform the original protein values into new values which have the goal distribution.…”

“…Both methods were shown to be effective in removing batch to batch variation in the datasets analysed. The observation that batch differences can affect cell subpopulations differently (Van Gassen et al, 2019) suggests that it will not be sufficient to apply a single batch adjustment to the measurements on all cells, as it is being done in BatchAdjust (Schuyler et al, 2019). However, a comparison of the two methods assessing the performance and limitations of each method has not been performed.…”