Table 4 Classification schema for Classier #3 (Method for predicting dual specificity kinases). HDTree Classifier #3 uses the outputs from HDTree Classifier #1 and HDTree Classifier #2 to distinguish between dual-specificity kinases, Ser/Thr kinases, and Tyr kinases. There are four possible labelings from the binary classifiers #1 and #2. 'Yes' or 'No' votes from Classifier #1 correspond to predictions of Ser/Thr or Tyr labels, respectively, for the protein. 'Yes' or 'No' votes from Classifier #2 correspond to predictions of Tyr or Ser/Thr labels. When both classifiers predict the protein to be Ser/Thr (that is, Classifier #1 votes 'Yes' and Classifier #2 votes 'No'), Classifier #3 labels the protein as "exclusively Ser/Thr" (and hence, not Tyr). Similarly, when both classifiers predict the protein to be Tyr, Classifier #3 labels the protein as "exclusively Tyr" (and hence not Ser/Thr). When both classifiers vote 'Yes' or when both vote 'No,' Classifier #3 labels the protein as having "Dual" catalytic activity. See Methods section for details on each classifier.

From: Exploring inconsistencies in genome-wide protein function annotations: a machine learning approach