Table 1.
Performance of classifiers.
| Model | Training cohort | Validation Cohort | P | ||||||
|---|---|---|---|---|---|---|---|---|---|
| AUC (95% CI) | Acc (%) | Sen (%) | Spe (%) | AUC (95% CI) | Acc (%) | Sen (%) | Spe (%) | ||
| LR_U | 0.977 (0.973–0.980) | 97.75 | 97.35 | 97.99 | 0.977 (0.972–0.983) | 97.80 | 97.51 | 97.97 | 0.83 |
| LDA_U | 0.966 (0.962–0.970) | 96.26 | 98.07 | 95.17 | 0.963 (0.957–0.969) | 95.83 | 98.07 | 94.48 | 0.36 |
| SVM_U | 0.977 (0.974–0.981) | 97.83 | 97.28 | 98.15 | 0.977 (0.971–0.982) | 97.77 | 97.27 | 98.06 | 0.88 |
| RF_U | 0.998 (0.997–0.999) | 98.15 | 97.65 | 98.45 | 0.976 (0.971–0.982) | 97.83 | 97.34 | 98.12 | 3e−14 |
| LR_H | 0.961 (0.956–0.966) | 96.30 | 95.07 | 97.17 | 0.965 (0.958–0.972) | 96.60 | 95.97 | 97.06 | 0.36 |
| LDA_H | 0.947 (0.941–0.952) | 94.68 | 94.49 | 94.81 | 0.938 (0.929–0.947) | 93.66 | 94.37 | 93.16 | 0.10 |
| SVM_H | 0.961 (0.957–0.966) | 96.36 | 94.96 | 97.35 | 0.965 (0.958–0.97) | 96.64 | 95.71 | 97.30 | 0.42 |
| RF_H | 0.998 (0.998–0.999) | 97.81 | 97.17 | 98.27 | 0.976 (0.970–0.981) | 97.58 | 96.67 | 98.25 | 1e−14 |
LR_U, LDA_U, SVM_U, and RF_U means the classifiers based on ribosome-protected fragment sequencing of U2OS cells using logistic regression, linear discriminant analysis, support vector machine, and random forest models, respectively. LR_H, LDA_H, SVM_H, and RF_H means the classifiers based on ribosome-protected fragment sequencing of HeLa cells using logistic regression, linear discriminant analysis, support vector machine, and random forest models, respectively. AUC, the area under ROC curve; Acc, accuracy; Sen, sensitivity; Spe, specificity; P. P value of the AUC difference between training and validation cohort.