Community Research and Development Information Service - CORDIS

ERC

MIXTURE Report Summary

Project ID: 336159
Funded under: FP7-IDEAS-ERC
Country: United Kingdom

Mid-Term Report Summary - MIXTURE (Synergistic Modelling of Molecular Effects via Chemical and Biological Data Integration)

Understanding, as well as predicting, the effect of compound combinations is of crucial importance, both for efficacy (such as drug combination therapies in cancer), as well to anticipate toxic side effects (both of drug-drug interactions, but also eg in consumer products). This project aims to understand the effect of compound combinations, based on the analysis of large experimental data sets in combination with suitable computational modelling methods.

To achieve this goal, data from several areas has been compiled in the first step, from medicinally relevant areas, namely from the areas of cancer, malaria, as well as the antibiotic area. In all of those datasets the effect of compound combinations has been experimentally measured – so for example the effect of two currently approved anti-cancer drugs was measured in combination against a particular cancer cell line, and so on. In the next step, chemical and biological descriptors (such as chemical structure, gene expression data from the biological system, etc.) were calculated in order to represent the data to the computer – with an associated output variable, which described the experimental effect of both compounds in combination. Finally, a machine learning algorithm was employed in order to model the current dataset at hand – and to make predictions about novel drug combinations. Those novel drug combinations were tested experimentally in all of the above areas, in order to estimate how well the model works in practice.

It was found that by using either chemical descriptors, and/or gene expression data (as well as some other biological descriptors), we were able to model the combination effects of compounds against cancer cell lines, the malaria parasite, as well as bacteria, better than random. In some cases (such as malaria) gene expression data significantly improved the prediction of the efficacy of drug combinations, over using chemical information alone. All of the above predictions have been tested in experiments with collaboration partners (such as the National Institutes of Health/NIH, Harvard University, as well as AstraZeneca), which gives us more confidence that they would also work for novel drugs, and also particular patient populations.

In the next step of the project we now aim to explore methods further, to widen the experimental validation, and to compile a ‘best practice’ guideline (as well as software) for modelling compound combinations. This will then also help other scientists, as well as ultimately the treatment of patients with combination therapies.

Reported by

THE CHANCELLOR, MASTERS AND SCHOLARS OF THE UNIVERSITY OF CAMBRIDGE
United Kingdom
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top