T cells post-COVID19

To identify T cell populations associated with restrictive lung disease post-COVID19
Conclusion:
Machine learning revealed marked CCR5+CD95+ CD8+ T-cell perturbations in mild-to-moderate lung disease, but attenuated T-cell responses in more severe disease.
Notes:
Please refer to the manuscript for the complete panel details and staining protocol. All spectral flow cytometry data were pre-processed by spectral unmixing with autofluorescence subtraction and spill-over correction. Fluorescence parameters were arcsinh-transformed with custom cofactors. Removal of dead cells, debris doublets and atypical events (antibody aggregates) was performed by expert gating. Similarly, expert gating was used to pre-gate for CD3+ lymphocytes. All data was dimensionality reduced using UMAP and clustered by FlowSOM in order to perform batch normalization using CytoNorm (OMIQ, Dotmatics, Boston, MA, USA). Two batch control samples were used in each experiment as internal controls; one to train batch normalization, and the other to test the validity of the normalization. Two antibodies, CD28-BV650 and PD1-BV785, were not included in batch 1, and were thus excluded from batch normalization. After normalization, MFI peaks of each fluorophore were inspected for alignment across batches, and batch control samples were inspected for the presence of non-overlapping regions between batches on dimensionalityreduction maps.