Fig. 1

An optimized signature for immune evaluation through introducing multi-regional sequencing data into signature construction. a Schematic illustration of the study design. b The distribution of the signature scores of multi-regional tumor samples from Renji (upper) and GSE136711 (lower) cohort classified by 20-gene signature. Tumors were divided into inflamed and non-inflamed classes based on tertiles. Gray nodes represent the tumor samples from the same patient classified into a concordant classification. Red nodes represent the tumor samples from the same patient classified into discordant classifications. c The proportion of concordant and discordant classification in Renji and GSE136711 cohort classified by 20-gene signature. A concordant tumor means that the multi-region samples in this tumor all belong to a single class (inflamed or non-inflamed class), while a discordant tumor means that multi-region samples in this tumor belong to both inflamed and non-inflamed class. d The distribution of the signature scores of multi-regional tumor samples from Renji and GSE136711 cohort classified by optimized 29-gene signature. e The proportion of concordant and discordant classification in Renji and GSE136711 cohort classified by 29-gene signature. f Expression pattern of genes in the 29-gene signature in Renji cohort. g Correlation analysis of signature scores between 20-gene signature and optimized 29-gene signature in Renji and GSE136711 cohort. h Heatmap describing the molecular and immune characteristics of inflamed and non-inflamed HCC classes determined by 29-gene signature or 20-gene signature in Renji cohort. i Schematic diagram depicting the finding of the current study