Figure 7. Associations between HLA-genotypes in COVID-19 patients and abundance of epitope specific CDR3 k-mers or meta-clonotypes.

(A) Beta-binomial regression coefficient estimates (x-axis) for participant genotype matching a hypothesized restricting HLA allele and negative log₁₀ false discovery rates (y-axis) for features developed from CD8+ TCRs activated by one of 16 HLA-restricted SARS-CoV-2 epitopes found in ORF1ab, ORF3a, nucleocapsid (N), and surface glycoprotein (S). Regression models included age, sex, and days post diagnosis as covariates (not shown). Positive HLA coefficient estimates correspond with greater abundance of the TCR feature in those patients expressing the restricting allele. (B) Distribution of false discovery rates by feature identification method (k-mer local, k-mer global, or meta-clonotype). Larger negative log₁₀-tranformed FDR values (y-axis) indicate more statistically significant associations. Local k-mer (e.g., FRTD) and global k-mer (e.g., SFRTD.YE) were identified using GLIPH2 (Huang et al., 2020) and were used to quantify counts of conforming TCRs in each bulk sequenced COVID-19 repertoire (see Method for details).