Fig. 1

Computational method for estimating the abundance of tumor-infiltrating immune cells. Tumor purity was estimated for each sample using DNA single-nucleotide polymorphism (SNP) array data (a). B allele frequency (BAF) is the frequency of a randomly selected parental allele. The logR ratio (LRR) is the log₂(Y/2), Y being the marker intensity in the SNP array. TCGA gene expression profiles were combined with reference immune cell expression data after batch effect removal (b). Informative genes with expression levels inversely correlated with tumor purity (Pearson’s r ≤ −0.2 and P value ≤ 0.05) are selected (c) and tested for immune signature enrichment (Fisher’s exact test) (d). In all 23 cancers informative genes are significantly enriched for immune signature. Diffuse large B-cell lymphoma (DLBC) has the lowest enrichment (odds ratio = 1.6, q = 0.0005, Fisher’s exact test). In this study, we estimate the abundance of six immune cell types (B cells, CD4 T cells, CD8 T cells, neutrophil, macrophage, and dendritic cells) using selected immune signature genes through constrained least squares fitting (e). Asterisks in d indicate event significance at a 1 % false discovery rate