Table 3.
Performance of a logistic regression algorithm for classifying genes into one of three Gene Ontology (GO) roles based on genes’ expression profiles across the RNA-seq samples. There are three GO roles, cellular component, molecular function, and biological process, and genes may correspond to one, two, or three of these roles. Performance is assessed by six measures: accuracy, precision, recall, F1 score, area under ROC curve, and area under precision-recall curve.
| Cellular Component | Molecular Function | Biological Process | |
|---|---|---|---|
| Accuracy | 75% | 83% | 86% |
| Precision | 75% | 84% | 86% |
| Recall | 100% | 100% | 100% |
| F1 score | 86% | 91% | 93% |
| Area under ROC curve | 70% | 68% | 66% |
| Area under PR curve | 88% | 91% | 93% |