Propagation on Molecular Interaction Networks: Prediction of Effective Drug Combinations and Biomarkers in Cancer Treatment.

BACKGROUND: Biomedical sciences use a variety of data sources on drug molecules, genes, proteins, diseases and scientific publications etc. This system can be best pictured as a giant data-network linked together by physical, functional, logical and similarity relationships. A new hypothesis or discovery can be considered as a new link that can be deduced from the existing connections. For instance, interactions of two pharmacons - if not already known - represent a testable novel hypothesis. Such implicit effects are especially important in complex diseases such as cancer.

METHODS: The method we applied was to test whether novel drug combinations or novel biomarkers can be predicted from a network of existing oncological databases. We start from the hypothesis that novel, implicit links can be discovered between the network neighborhoods of data items.

RESULTS: We showed that the overlap of network neighborhoods is strongly correlated with the pairwise interaction strength of two pharmacons used in cancer therapy, and it is also well correlated with clinical data. In a second case study we employed this strategy to the discovery of novel biomarkers based on text analysis. In 2012 we prioritized 10 potential biomarkers for ovarian cancers, 2 of which were in fact described as such in the subsequent years.

CONCLUSION: The strategy seems to hold promises for prioritizing new drug combinations or new biomarkers for experimental testing. Its use is naturally limited by the sparsity and the quality of experimental data, however both of these aspects are expected to improve given the development of current databases.