Figure 2 from Inferring Characteristics of the Tumor Immune Microenvironment of Patients with HNSCC from Single-Cell Transcriptomics of Peripheral Blood

The ICFs of major immune cell types in the TME can be predicted from the blood. A, Correlation matrix of ICFs between the TME and the blood. For immune cell types in the TME (columns), the cell types in the blood (rows) with the highest correlations are marked in black boxes. Pearson correlation values shown are mean ± SD over 1,000-replicate bootstrapping. B, Distribution of correlations between the model-predicted ICFs and the true TME ICFs are shown on the test sets across 1,000 replicates. C, Correlation between the model-predicted TME ICF and the measured TME ICF for each cell type on a small, independent validation dataset (n = 5). Cycling T cells are not studied as they are not present in all tumor tissue samples; monocytes and DC are excluded as their pertaining predictive clinical information is missing. r denotes the Pearson correlation coefficient, and ρ denotes the Spearman correlation coefficient. Statistical notation: ***, FDR < 0.001; **, FDR < 0.01; *, FDR < 0.05. FDRs are calculated using Benjamini–Hochberg correction. Alc., alcohol; info., information; Tob., tobacco.