Fig. 5

Using LASSO regression analysis to construct the most predictable TRRGs signature. A LASSO coefficients distribution of 206 significantly prognostic TRRGs. B Tuning parameter (λ) selection cross‐validation error curve. Vertical lines were drawn at the optimal values. C Regression coefficient corresponding to the 20 screened TRRGs are shown. A larger absolute value of the coefficient represents a higher correlation. D Random Forest of 20 TRRGs. E Heatmap of the 20 TRRGs between remission and non-remission groups. F-G Different gene expression between remission and non-remission groups