Skip to main content
. 2022 Apr 25;12:861601. doi: 10.3389/fonc.2022.861601

Figure 1.

Figure 1

Construction of prognosis model. Thirty-two genes from the TGF-β family were subjected to univariate Cox regression analysis. (A) Seven TGF-β family members (including TGF-β1) were identified as independent prognostic factors for hepatocellular carcinoma (HCC) using survival, survminer, and forestplot packages. Red indicates positive correlation while green indicates negative correlation. p < 0.05 suggests statistical significance. LASSO regression analysis was applied to reduce model complexity and prevent over-fitting, with the optimal lambda values shown in (B) As indicated in (C), after analysis using glmnet and survival packages, each curve in the LASSO regression represents a gene, with the coefficients shown in the y-axis. The numbers on the top of (B, C) indicate how many genes are left in the model when Lambda is set to a certain value. The coefficients from top to bottom in (C) correspond to the respective genes as follows: GDF6 (0.41), GDF10 (0.36), LEFTY2 (0.34), BMP6 (0.17), BMP2 (0.16), TGFB1 (0.049), and GDF7 (−0.41).