Fine tuning a logical model of cancer cells to predict drug synergies: combining manual curation and automated parameterization

Abstract

Treatment with combinations of drugs carries great promise for personalized therapy for a variety of diseases. We have previously shown that synergistic combinations of cancer signaling inhibitors can be identified based on a logical framework, by manual model definition. We now demonstrate how automated adjustments of model topology and logic equations both can greatly reduce the workload traditionally associated with logical model optimization. Our methodology allows the exploration of larger model ensembles that all obey a set of observations, while being less restrained for parts of the model where parameterization is not guided by biological data. We benchmark the synergy prediction performance of our logical models in a dataset of 153 targeted drug combinations. We show that well-performing manual models faithfully represent measured biomarker data and that their performance can be outmatched by automated parameterization using a genetic algorithm. Whereas the predictive performance of a curated model is strongly affected by simulated curation errors, data-guided deletion of a small subset of regulatory model edges can significantly improve prediction quality. With correct topology we find evidence of some tolerance to simulated errors in the biomarker calibration data, yet performance decreases with reduced data quality. Moreover, we show that predictive logical models are valuable for proposing mechanisms underpinning observed synergies. With our framework we predict the synergy of joint inhibition of PI3K and TAK1, and further substantiate this prediction with observations in cancer cell cultures and in xenograft experiments.