Research Paper Volume 16, Issue 12 pp 10299—10320
Mining key circadian biomarkers for major depressive disorder by integrating bioinformatics and machine learning
Figure 5. Screening of key MDD-related CRGs using machine learning methods. (A) Optimal λ selection for LASSO regression model. (B) LASSO coefficient profiles of the 21 feature genes. (C) Top 30 feature importance from random forest algorithm. (D) Top 30 important genes ranked by random forest algorithm. (E) Accuracy rate plot of SVM-RFE model. (F) Error rate plot of SVM-RFE model. (G) Top 22 genes with lowest error rate ranked by SVM-RFE. (H) Venn diagram for screening of the 4 key CRGs (ABCC1, APP, HK2, and RORA) by integrating LASSO, SVM-RFE and Random Forest algorithms.