Figure 1.

A set of 307 PTC-specific IRGs were identified. Cumulative proportion curves showed that the immune scores were continuously distributed at the significantly higher side compared with stromal scores in both (a) xCell and (b) ESTIMATE outputs, indicating the predominant role of immune infiltration in PTC tumor microenvironment (TME). (c) A correlation network reflects the comprehensive relationships among different immune cell types involved in PTC TME. (d) The weighted gene co-expression network analysis (WGCNA) was performed with the IRGs expression matrix of 94 samples and their sample category (tumor or normal thyroid) to construct a scale-free co-expression network. (e) Three gene modules were generated, and a blue module exhibited the highest correlation with sample category (|r| = 0.9, P = 4e-35) and was considered as “PTC-specific module”. (f) Limma algorithm was used to identify a total of 492 DE-IRGs between PTC and normal thyroid samples with a filtering threshold of FDR Q value less than 0.01. (g) 307 overlapping genes in the intersection of “PTC-specific module” and “DE-IRGs” were considered as “PTC-specific IRGs”. DE-IRGs: differentially expressed immune-related genes. (h) PCA analysis demonstrated that PTC and normal thyroid samples were clearly separated as two distinct groups in the discovery cohort with the 307 PTC-specific IRGs expression matrix. (i-k) Three datasets (GSE60542, GSE3467, GSE3678) were used to validate the discriminative capacity of the 307 PTC-specific IRGs, and the dissimilarity between different sample categories was visualized in PCA analysis