Cancer is a very difficult disease to treat. Traditional treatments (surgery, chemotherapy, radiation) are often risky and lead to complications. Newer treatments that can potentially be very selective and with fewer side effects (targeted therapies) must be tailored to each patient, and are subject to become unhelpful due to development of resistance within the tumour. Regardless of the treatment type, there is a constant need to monitor the patient through various medical examinations. Oncologists follow on these examinations and decide whether and how to continue with the treatment. Such decisions are based on human intuition – the oncologist must be able to analyse a large body of data and estimate the likelihood of success for different alternatives.
Through integration of available data from the medical literature and/or omics experiments, we developed a software that identifies critical nodes in a biological interaction network for efficient therapeutic treatment and consequently helps oncologists to integrate patient data. Especially in the case of cancer resistance to therapies, tumours escape treatment by alternative signalling routes. The identification of such escape routes are crucial for preventing tumour development.
Currently available computer programs that simulate biological networks generally belong to one of two types. The first uses only binary data (e.g., can a certain disease marker be observed in cancer cells – Yes/No). The second use very detailed information (e.g., measurements of the marker’s concentration in a given tissue) that is often not available or hard to obtain. Our technology is unique in its ability to handle a third and more common type of data, where the available information is sometimes quite accurate and in other times missing or given only as a range of values.
The technology is based on analysis and solution of kinetic equations by modern tools. The novelty lies in our ability to build these equations at run time (abstraction) so that the user does not need to be aware of the procedure. Moreover, we have designed our program to be highly parallelizable and thus suited for today’s machines that are typically multicore, and to future generations of personal computers and mobile devices that are expected to include hundreds of computing cores per machine.
Buetti-Dinh A, Pivkin IV, Friedman R.
Characterising signal transduction networks is fundamental to our understanding of biology. However, redundancy and different types of feedback mechanisms make it difficult to understand how variations of the network components contribute to a biological process. In silico modelling of signalling interactions therefore becomes increasingly useful for the development of successful therapeutic approaches. Unfortunately, quantitative information cannot be obtained for all of the proteins or complexes that comprise the network, which limits the usability of computational models. We developed a flexible computational framework for the analysis of biological signalling networks. We demonstrate our approach by studying the mechanism of metastasis promotion by the S100A4 protein, and suggest therapeutic strategies. The advantage of the proposed method is that only limited information (interaction type between species) is required to set up a steady-state network model. This permits a straightforward integration of experimental information where the lack of details are compensated by efficient sampling of the parameter space. We investigated regulatory properties of the S100A4 network and the role of different key components. The results show that S100A4 enhances the activity of matrix metalloproteinases (MMPs), causing higher cell dissociation. Moreover, it leads to an increased stability of the pathological state. Thus, avoiding metastasis in S100A4-expressing tumours requires multiple target inhibition. Moreover, the analysis could explain the previous failure of MMP inhibitors in clinical trials. Finally, our method is applicable to a wide range of biological questions that can be represented as directional networks.Go To Mol Biosyst
Global Medical Discovery Feature