The biological determinants underlying the range of COVID-19 clinical manifestations are not fully understood. Here, over 1400 plasma proteins and 2600 single-cell immune features comprising cell phenotype, endogenous signaling activity, and signaling responses to inflammatory ligands are cross-sectionally assessed in peripheral blood from 97 patients with mild, moderate, and severe COVID-19 and 40 uninfected patients. Using an integrated computational approach to analyze the combined plasma and single-cell proteomic data, we identify and independently validate a multivariate model classifying COVID-19 severity (multi-class AUC_training = 0.799, p-value = 4.2e-6; multi-class AUC_validation = 0.773, p-value = 7.7e-6). Examination of informative model features reveals novel biological signatures of COVID-19 severity, including the dysregulation of JAK/STAT, MAPK/mTOR, and NF-κB immune signaling networks in addition to recapitulating known hallmarks of COVID-19. These results provide a set of early determinants of COVID-19 severity that may point to therapeutic targets for prevention and/or treatment of COVID-19 progression.

Mass cytometry:

Blood was collected in EDTA tubes. Anticoagulated blood was processed into peripheral blood mononuclear cells (PBMC) by density gradient centrifugation using published methods (Syed et al., 2014). PBMC were stored in 10% DMSO and frozen in liquid nitrogen until thawing and staining. 
Cryopreserved PBMC were quickly thawed, washed two times with supplemented medium, and rested for 1h at 37°C in RPMI 1640 medium supplemented with 10% fetal bovine serum, 1% Penicillin-Streptomycin, and 1% L-Glutamine. PBMC were counted and checked for viability. 0.5-1x10⁶ cells were either stimulated with lipopolysaccharide (LPS; 1?g/ml) and CL097 (TLR7/8 agonist; 1?g/ml), interleukin-2 (IL-2), IL-4, IL-6 and interferon-? (IFN?; all 100 ng/ml), a cocktail of phorbol 12-myristate 13-acetate (PMA), ionomycin, brefeldin A and monensin (1x; PI cocktail), or left unstimulated for 15 minutes at 37°C. After stimulation, samples were fixed with Proteomic Stabilizer (SmartTube) and stored at -80°C until further processing for mass cytometry analysis. 

Plasma proteomics:

Plasma was isolated from EDTA-collected blood within 4hrs of collection centrifuging tubes at 500 x g for 10 minutes at room temperature (RT). Plasma was transferred into fresh conical tubes and centrifuged again (500 x g, 10 minutes, RT) before aliquoting into 500 µL cryovials and transferred to -80°C for long-term storage. 
Plasma protein levels were quantified using Olink multiplex proximity extension assay (PEA) panels (Olink Proteomics; www.olink.com) according to the manufacturer’s instructions. Plasma was treated with 1% Triton X-100 for 2 hours at room temperature to inactivate any potential virus, before shipping to Olink Proteomics.
The raw expression values obtained with the Olink assay are provided in the arbitrary unit Normalized Protein Expression (NPX), where high NPX values represent high protein concentration. Values were log2-transformed to account for heteroskedasticity. Proteins close to the limit of detection are flagged in the raw data.

The data files contain:
- FCS files of mass cytometry measurements
- PatientCharacteristics: file containing clinical information regarding the patients. Column "Cohort" indicates which samples were used for training and which for validation.
- Penalization matrix: penalization matrix used to penalize the mass cytometry data
- Olink_NPXvalues: data from the Olink plasma proteomic assay. Proteomic values are provided in the arbitrary unit NPX
- CyTOF_FrequencySignaling_preprocessed: Frequencies and processed phosphosignaling responses of the mass cytometry dataset
- Correlation_matrix: Absolute spearman correlation coefficients of the chord diagram depicted in Figure 3D.

Further information and requests for resources and reagents should be directed to an will be fulfilled by the lead contact, Brice Gaudillière (gbrice@stanford.edu)