Fig. 2

Construction of the prognostic model based on the risk score. a LASSO coefficient profiles of the 2 survival-related mRNAs. Each curve corresponds to a gene. It shows the path of its coefficient against the L1-norm of the whole coefficient vector at various λ values. The vertical line is drawn at the value λ = 0.2828604 chosen by leave-one-out cross-validation. Two genes (ACAP1 and RASGRP1) intersecting with the vertical line were chosen to build the final model. b Partial likelihood deviance for the LASSO coefficient profiles. The red dotted line stands for the cross-validation curve, error bars represent the upper and lower standard deviation curves along the λ sequence. The left vertical line shows the optimal λ value at which the minimum mean squared error is achieved and the corresponding genes. The right vertical line is for the most regularized model whose mean squared error is within 1 standard error of the minimal. It is indicated that the genes identified by optimal λ are the simplest model with the best performance. In a, b, the axis above indicates the number of genes involved in the LASSO model. c The expression levels of ACAP1 and RASGRP1 between low and high-risk groups in the training and validation set. The high-risk group (risk score ≥ 0.6715958) had significantly lower proportions of ACAP1 and RASGRP1 than the low-risk group (risk score < 0.6715958) in both datasets. All p values were corrected by the Bonferroni method. (Wilcoxon rank-sum test, *p-value < 0.05, ****p-value < 0.0001)