Raw and processed (filtered and annotated) scRNAseq data

Single cell RNA-seq data generated and reported as part of the manuscript entitled "<strong>Dissecting the mechanisms underlying the Cytokine Release Syndrome (CRS) mediated by T Cell Bispecific Antibodies</strong>" by Leclercq-Cohen et al 2023. Raw and processed (filtered and annotated) data are provided as AnnData objects which can be directly ingested to reproduce the findings of the paper or for ab initio data reuse: 1- raw.zip provides concatenated raw/unfiltered counts for the 20 samples in the standard Market Exchange Format (MEX) format. 2- 230330_sw_besca2_LowFil_raw.h5ad contains filtered cells and raw counts in the HDF5 format. 3- 221124_sw_besca2_LowFil.annotated.h5ad contains filtered cells and log normalized counts, along with cell type annotation in the HDF5 format. ############################## <strong>scRNAseq data generation:</strong> Whole blood from 4 donors was treated with 0.2 μg/mL CD20-TCB, or incubated in the absence of CD20- TCB. At baseline (before addition of TCB) and assay endpoints (2, 4, 6, and 20 hrs), blood was collected for total leukocyte isolation using EasySepTM red blood cell depletion reagent (Stemcell). Briefly, cells were counted and processed for single cell RNA sequencing using the BD Rhapsody platform. To load several samples on a single BD Rhapsody cartridge, sample cells were labelled with sample tags (BD Human Single-Cell Multiplexing Kit) following the manufacturer’s protocol prior to pooling. Briefly, 1x106 cells from each sample were re-suspended in 180 μL FBS Stain Buffer (BD, PharMingen) and sample tags were added to the respective samples and incubated for 20 min at RT. After incubation, 2 successive washes were performed by addition of 2 mL stain buffer and centrifugation for 5 min at 300 g. Cells were then re- suspended in 620 μL cold BD Sample Buffer, stained with 3.1 μL of both 2 mM Calcein AM (Thermo Fisher Scientific) and 0.3 mM Draq7 (BD Biosciences) and finally counted on the BD Rhapsody scanner. Samples were then diluted and/or pooled equally in 650 μL cold BD Sample Buffer. The BD Rhapsody cartridges were then loaded with up to 40 000 – 50 000 cells. Single cells were isolated using Single-Cell Capture and cDNA Synthesis with the BD Rhapsody Express Single-Cell Analysis System according to the manufacturer’s recommendations (BD Biosciences). cDNA libraries were prepared using the Whole Transcriptome Analysis Amplification Kit following the BD Rhapsody System mRNA Whole Transcriptome Analysis (WTA) and Sample Tag Library Preparation Protocol (BD Biosciences). Indexed WTA and sample tags libraries were quantified and quality controlled on the Qubit Fluorometer using the Qubit dsDNA HS Assay, and on the Agilent 2100 Bioanalyzer system using the Agilent High Sensitivity DNA Kit. Sequencing was performed on a Novaseq 6000 (Illumina) in paired-end mode (64-8- 58) with Novaseq6000 S2 v1 or Novaseq6000 SP v1.5 reagents kits (100 cycles). <strong>scRNAseq data analysis:</strong> Sequencing data was processed using the BD Rhapsody Analysis pipeline (v 1.0 https://www.bd.com/documents/guides/user-guides/GMX_BD-Rhapsody-genomics- informatics_UG_EN.pdf) on the Seven Bridges Genomics platform. Briefly, read pairs with low sequencing quality were first removed and the cell label and UMI identified for further quality check and filtering. Valid reads were then mapped to the human reference genome (GRCh38-PhiX-gencodev29) using the aligner Bowtie2 v2.2.9, and reads with the same cell label, same UMI sequence and same gene were collapsed into a single raw molecule while undergoing further error correction and quality checks. Cell labels were filtered with a multi-step algorithm to distinguish those associated with putative cells from those associated with noise. After determining the putative cells, each cell was assigned to the sample of origin through the sample tag (only for cartridges with multiplex loading). Finally, the single-cell gene expression matrices were generated and a metrics summary was provided. After pre-processing with BD’s pipeline, the count matrices and metadata of each sample were aggregated into a single adata object and loaded into the besca v2.3 pipeline for the single cell RNA sequencing analysis (43). First, we filtered low quality cells with less than 200 genes, less than 500 counts or more than 30% of mitochondrial reads. This permissive filtering was used in order to preserve the neutrophils. We further excluded potential multiplets (cells with more than 5,000 genes or 20,000 counts), and genes expressed in less than 30 cells. Normalization, log-transformed UMI counts per 10,000 reads [log(CP10K+1)], was applied before downstream analysis. After normalization, technical variance was removed by regressing out the effects of total UMI counts and percentage of mitochondrial reads, and gene expression was scaled. The 2,507 most variable genes (having a minimum mean expression of 0.0125, a maximum mean expression of 3 and a minimum dispersion of 0.5) were used for principal component analysis. Finally, the first 50 PCs were used as input for calculating the 10 nearest neighbours and the neighbourhood graph was then embedded into the two-dimensional space using the UMAP algorithm at a resolution of 2. Cell type annotation was performed using the Sig-annot semi-automated besca module, which is a signature- based hierarchical cell annotation method. The used signatures, configuration and nomenclature files can be found at https://github.com/bedapub/besca/tree/master/besca/datasets. For more details, please refer to the publication.