Fig. 2: Myeloid compositional changes and gene expression alterations in cardiac inflammation.

a, UMAP embedding delineated 16 myeloid cell states. b, Upper panel, mean abundance (%) of myeloid cell states in control left ventricles (n = 18). Lower panel, proportional changes of myeloid cell states in the four patient groups (Non-COVID-19: n = 8, Post-COVID-19: n = 10, Post-Vaccination: n = 4). Color scale: red (increase) and blue (decrease). P values are indicated for significant proportional changes (FDR ≤ 0.05). P values were calculated using the two-sided t-test based on CLR-transformed values with Benjamini–Hochberg correction. For MIS-C, significance was not calculated due to low sample size (n = 2). c, Condition-split UMAP showing compositional shifts in myeloid cell states across patient groups and controls. d,e, Box plots showing distribution of Lyve1^hiMHCII^lo and Lyve1^loMHCII^hi (d) and CD16⁺ and VCAN⁺ monocytes (e) across control and disease groups (control: n = 18, Non-COVID-19: n = 8, Post-COVID-19: n = 10, Post-Vaccination: n = 4). Boxes depict the interquartile range (IQR); horizontal bars indicate the median; whiskers extend to 1.5× IQR; and dots show the value of each patient. P values are indicated for significant proportional changes, FDR < 0.05. P value calculations are as described in b. f, Dot plots showing differential expression of selected IFNγ response and MHC-II genes in patient groups relative to control across Lyve1^hiMHCII^lo and Lyve1^loMHCII^hi (control: n = 18, Non-COVID-19: n = 8, Post-COVID-19: n = 10, Post-Vaccination: n = 4). Dot colors indicate log₂-transformed FCs (log₂FCs). Dot sizes indicate absolute log₂FC. Black circles indicate significance (FDR ≤ 0.05). P values were calculated using the quasi-likelihood F-test and were adjusted for multiple testing (Benjamini–Hochberg). Genes are ordered alphabetically.

Source data