Fig. 4

The GSAGI predicts the immune characteristics of LUAD. A ESTIMATE algorithm evaluates the immune score of patients in the high-risk and low-risk groups in the TCGA-LUAD training set. B ESTIMATE algorithm evaluates the tumor purity of patients in the TCGA-LUAD training set. C The CIBERSORT algorithm assesses the differences in the degree of infiltration of B cells naive, NK cells activated, T cells CD4 naive, and Macrophages M1 between the high-risk and low-risk groups in the TCGA-LUAD set. **P < 0.01 and ****P < 0.0001. D Four algorithms in the TIMER database assess the differences in the degree of infiltration of B cells between the high-risk and low-risk groups in the TCGA-LUAD training set. *P < 0.05 and ***P < 0.001. E Positive correlations between TMB levels and the expression of ANLN, RHOV, KRT6A, and KLRG2, and some negative correlations with SIGLEC6 expression in LUAD patients. F-G Correlation between risk scores of LUAD patients and the expression of 17 chemokines and their receptors (F) and the expression of 7 immune checkpoint molecules (G). The colors represent Pearson correlation coefficients, and the sizes of the ellipses represent the P-values. H Two-way bar graphs show IPS for patients in the high-risk and low-risk groups in the TCGA-LUAD training set. ***P < 0.001. I Comparison of TIDE scores between patients in the high-risk and low-risk groups in the TCGA-LUAD training set. J The percentage bar graph compares the different response statuses of patients receiving immunotherapy in the high-risk and low-risk groups in the TCGA-LUAD training set. ***P < 0.001. Red indicates that the patient responded to ICI treatment, blue indicates non-response