Development of a Comprehensive Antibody Staining Database using a Standardized Analytics Pipeline

Abstract

15Large-scale immune monitoring experiments (such as clinical trials) are a promising direction for 16 biomarker discovery and responder stratification in immunotherapy. Mass cytometry is one of the 17 tools in the immune monitoring arsenal. We propose a standardized workflow for the acquisition and 18 analysis of large-scale mass cytometry experiments. The workflow includes two-tiered barcoding, a 19 broad lyophilized panel, and the incorporation of a fully automated, cloud-based analysis platform. 20We applied the workflow to a large antibody staining screen using the LEGENDScreen kit, resulting 21 in single-cell data for 350 antibodies over 71 profiling subsets. The screen recapitulates many known 22 trends in the immune system and reveals potential markers for delineating MAIT cells. Additionally, 23we examine the effect of fixation on staining intensity and identify several markers where fixation 24 leads to either gain or loss of signal. The standardized workflow can be seamlessly integrated into 25 existing trials. Finally, the antibody staining data set is available as an online resource for researchers 26 who are designing mass cytometry experiments in suspension and tissue. 27

Stabilizing Antibody Cocktails for Mass Cytometry

“…As an alternative to frozen antibody cocktails, lyophilized cocktails have been used in mass cytometry . However, the freezing approach has the advantage that it does not depend on a propriety platform, reagents do not require shipping, and adaptions to the antibody panel design can be quickly implemented.…”

Data processing workflow for large-scale immune monitoring studies by mass cytometry

“…Thus, it is advisable to verify the quality and channel intensities of the different aliquots acquired in each acquisition batch. To do this, we took advantage of AOF algorithm [23] that detects potential staining problems [24] .…”

“…3 B and scaled to obtain Scaled AOF scores as illustrated in Fig. 4 A and described in [24] . These scores were summed for each aliquot, and the Quality AOF score was obtained for each fcs file as shown in Fig.…”

“…If the user decides not to split the data, we advise to carefully inspect the files with high quality scores and markers with high AOF score across time. It can happen that upon the long acquisition only a portion of the data is affected, thus some good quality part of the data can be used for further analysis, as shown before [24] .…”

“…The FlowSOM was built using a 10x10 grid and 10 metaclusters. We calculated AOF scores per aliquot and batch using greedyCytometryAof() function, with the default parameters from the cytutils package ( https://github.com/ismms-himc/cytutils ), and Scaled and Quality AOF using the formula presented in [24] . Next, we estimated the mean of Quality AOF scores taking into consideration the scores calculated for all the fcs files.…”