Fig. 2

The composition of the tumor microenvironment reflects specific immunogenic features. a Immunophenoscore signatures of the TME-associated groups. Median values per group of signatures of suppressor cells (SC), checkpoint molecules (CP), MHC molecules and effector cells (EC) were Z-score transformed and compared. b Neoepitope counts (in log scale) predicted for each patient’s HLA-I allele. c Log values of somatic mutation burden per genome megabase (Mb). Only single nucleotide variants called by VarScan2 and at least one additional tool (MuSE or SomaticSniper) were considered. d Number of neoepitopes per group considering samples with/without deleterious mutations in genes belonging to the antigen processing and presentation pathway. e HLA-I binding affinity distribution of predicted neoepitopes among groups. For each sample, we used the median HLA-I binding affinity. Statistical analysis in b, c and e were performed by MW test: *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. KD—dissociation constant for HLA-I binding in nM