Figure 3.

Sequence, chromatin, and DNA-binding proteins classify ORC-binding sites. (A) SVM performance was gauged by the ROC curve resulting from separately using sequence features, chromatin mark features, binding protein features, or a combination of all three. In each case, the SVM was trained using 10-fold cross validation on three chromosome arms (2L, 3L, and 3R) and tested on a fourth chromosome arm (2R). (B) The importance of individual features was determined by plotting the F-score as a function of class proximity represented here as a t-statistic.