1 to 0. 8 when genomic characterizations are used to predict the drug sensitivities in the CCLE study. In comparison, our approach based on sensitivity data on training set of drugs and drug protein interaction information produced correlation coefficients 0. 92 for both leave one out and 10 fold cross validation approaches for error estimation. It should be noted that the sensitivity unfortunately prediction is per formed in a continuous manner, not discretely, and thus effective dosage levels can be inferred from the predic tions made from the TIM. This shows that the TIM frame work is capable of predicting the sensitivity to anti cancer targeted drugs outside the training set, and as such is viable as a basis for a solution to the complicated problem of sensitivity prediction.

In addition, we tested the TIM framework using syn thetic data generated from a subsection of a human cancer pathway taken from the KEGG database. Here, the objective is to show that the proposed TIM method gener ates models that highly represent the Inhibitors,Modulators,Libraries underlying biological network which was sampled via synthetic drug pertur bation data. This experiment replicates in synthesis the actual biological experiments performed at the Keller lab oratory at OHSU. To utilize the TIM algorithm, a panel of 60 targeted drugs pulled from a library of 1000 is used as a training panel to sample the randomly generated network. Additionally, a panel Inhibitors,Modulators,Libraries of 40 drugs is drawn from the library to serve as a test panel. The training panel and the testing panel have no drugs in common.

Each of the 60 train ing drugs is applied to the network, and the sensitivity for each drug is recorded. The generated TIM is then sam pled using the test panel which determines the predicted sensitivities of the test panel. Dacomitinib The synthetic experiments were performed for 40 randomly generated cancer sub networks for each of n 6, 10 active targets in the network. The active targets are those which, when inhib ited, may have some effect on the cancer downstream. To more accurately mimic Inhibitors,Modulators,Libraries the Boolean nature of the biolog ical networks, a drug which does not satisfy any of the Boolean network equations will have sensitivity 0, a drug which satisfies at least one network equation will have sen sitivity 1. The inhibition profile of the test drugs is used Inhibitors,Modulators,Libraries to predict the sensitivity of the new drug.

The average number of correctly predicted drugs for each n is reported in Table 7. This synthetic modeling approach generally produces respectable levels of accuracy, with accuracies ranging from 89% to 99%. 60 drugs for training mimics useful site the drug screen setup used by our collaborators and testing 20 drugs for predicted sensitivity approximates a sec ondary drug screen to pinpoint optimal therapies. The performance of the synthetic data shows fairly high relia bility of the predictions made by the TIM approach.