I need to evaluate the accuracy of predicted interactome of an insect species which I constructed by interolog method. I read that I can use ROCR package for it. I got a gold standard positive PPI set and negative set and have a scoring scheme to classify my predictions as high confidence and low confidence. Could anyone guide me to proceed ahead for the ROC analysis?
Select several thresholds for your ROC analysis; for ex., say n=5 thresholds. For each threshold, perform your algorithm on the test set (positive controls and negative controls) and determine the number of true positives and true negatives, and false positives and false negatives. Calculate the true positive rate (TPR; sensitivity) and the false positive (FPR; 1-specificity) rates for each point. Then plot FPR and TPR on XY axes. This is your ROC curve (parabolic) , and the area under it is an indication of the predictive power of your classification algorithm.
- Badges
- Science topic
- Similar topics
- Bioinformatics
- Bioinformatics and Computational Biology