BioHealth – From Computational Biology and Medical Informatics to Health and Wellbeing

Research area

During recent decades, computational biology, bioinformatics, medical, and health informatics have claimed and successfully established their status as scientific disciplines. They are considered highly relevant for the wellbeing of modern information society. Traditionally, however, these disciplines have focused on different research topics and they have not maintained a dialogue resulting in satisfactory interaction from the point of view of health and wellbeing, the ultimate goals of both individuals and society at large. This TUCS Research Programme aims at establishing such a dialogue by setting up joint activities for the scientific disciplines as well as supporting interaction between academia and the rest of the society.

Research goal

The scope addressed in this initiative stretches over three tiers of varying levels of abstraction. The lowest level is the molecular level, where the research focus is on the interactions of various biomolecules, drugs, and conditions. The next layer is focused on medical and health related phenomena concerning a certain individual. Finally, the uppermost layer is concerned with societal aspects of ICT in health and wellbeing, such as health information systems. Currently, we witness a strong trend towards integrating these aspects, resulting in highly personalized medical treatments and active participation in the preventive, diagnostic, curative, and other processes of the health care system by empowered individuals.

Programme leader

Tapio Salakoski and Ion Petre

Participating TUCS Research Units

External partners

Department of Nursing Science at University of Turku (Sanna Salanterä)
Genetic interactions and network medicine at FIMM (Tero Aittokallio)
Turku Proteomics Facility (Garry Corthals)
Bioinformatics Core Facility at Turku Centre for Biotechnology (Attila Gyenesei)
Bioenergy group at Turku Centre for Biotechnology (Patrik Jones)

Steering group

Tero Aittokallio Barbro Back Ralph-Johan Back Garry Corthals Attila Gyenesei Jukka Heikkonen Patrik Jones Olli Nevalainen Kalle Parvinen Ion Petre Tapio Salakoski Sanna Salanterä Reima Suomi

Activities

14.5.2013: guest talk: Professor John D. Nagy (Scottsdale Community College and Arizona State University, USA)

Evolution of Proliferation and the Angiogenic Switch in Phenotypically Diverse Tumors

Cancer therapy is most commonly complicated by an incipient treatment resistance that eventually overwhelms all attempts at clinical intervention. This process is widely recognized as an evolutionary phenomenon in which treatment-resistant cells are favored by natural selection in the environment generated by therapy. Therefore, clinical strategies that harness natural selection – in contrast to current protocols that either ignore it or attempt to avoid it – may offer improved outcomes. One intriguing possibility in this direction is to force the tumor to evolve its own tumor, or ``hypertumor," generated by runaway selection for either extreme vascular hypo- or hyperplasia. This phenomenon is predicted by models of tumor angiogenesis studied with the techniques of adaptive dynamics. However, such techniques are predicated on two key assumptions: (i) no more than two distinct clones or evolutionary strategies can exist in the tumor at any given time; and (ii) mutations only have small effects on the evolutionary strategy. Here we relax these assumptions at the expense of mathematical tractability and show, using a stochastic simulation of angiogenesis evolution, that the predictions of the adaptive dynamics model do not depend on the adaptive dynamics assumptions. In particular, the evolutionary stable strategy for angiogenesis remains an evolutionary repeller. Furthermore we implement the use of the Shannon diversity index to measure convergence towards an evolutionary endpoint in phenotypically diverse tumors. Future numerical investigations will be directed towards identifying evolutionary stable strategies in a two dimensional strategy space where both proliferation and angiogenic commitment strategies are allowed to evolve. These predictions may allow better understanding of tumor progression through evolutionary mechanisms allowing for therapeutics that better utilize natural selection.

Host: Kalle Parvinen
12.10.2012: guest talk: Dr. Vladimir Rogojin (University of Helsinki)

Cancer Related Studies in Cellular Signalling Netowrks: Identifying Essential Nodes Via Feedback Loops.

Feedback loops are an inherent part of cell signalling networks and are responsible in maintaining robustness in cells. We present here a score that indicates the importance of the nodes in a signalling network based on their participation in feedback loops. Due to the high number of feedback loops in a signalling network, we have employed a graph compression technique to reduce the size and the complexity of the network.

We used two network models to generate random networks and compared our feedback loop centrality score with some classical node centrality measures. Our results indicate that the feedback loop based score identifies nodes that would have left undetected with the existing centrality measures.

We considered gene expression data from 384 glioblastoma multiforme patients and constructed a comprehensive intra-cellular signalling network based on genes with the survival effect. Interestingly, a survival associated gene STAT3 has the highest feedback loop score and participates in 95% of all the feedback loops in our network, which may explain its central role in the highly aggressive nature of glioblastoma multiforme cells.

Click here for the slides of the talk.

Host: Ion Petre
18.6.2012: guest talk: Dr. David Safranek (Masaryk University Brno, Czech Republic)

Parameter Indentification in Biochemical Dynamic Systems: The Model Checking Approach

Kinetic models reconstructing dynamics of biological processes, modeled at several levels of abstraction starting with purely qualitative discrete models and ending with quantitative deterministic or even stochastic ones, provide the corner-stone tool in systems biology. A very important issue is that of parameter uncertainty. Parameters appear in dynamical models of all kinds. Examples are kinetic rate coefficients driving the dynamics of quantitative models or kinetic logic parameters which set the dynamics of discrete models. In all cases, it is difficult to find robust parametrizations of models. Additionally to experimental data, known static and dynamic constraints significantly limit the possible model behaviour and help in searching for good and realistic parametrizations. In this talk we will present an approach to parameter identification based on model checking. Emphasis will be given to efficient and automatable methods. Presented techniques will target discrete and continuous models supplied with linear temporal specifications (constraints).

Host: Ion Petre
7.6.2012: guest talk: Dr. Serghey Verlan (University of Paris 12, France)

Applications of the Theory of Formal Languages for the Design of Fast Hardware Implementations of Parallel Multiset Rewriting

In this talk we present the design of a fast hardware simulator for parallel multiset rewriting using the field-programmable gate array (FPGA) technology. The simulator is non-deterministic and it uses a constant time procedure to choose one of the computational paths. The obtained strategy is equitable and it is based on a pre-computation of all possible rule applications. This pre-computation is obtained by using the representation of all possible multisets of rules' applications as context-free languages. Then using the Schutzenberger technique for the computation of the generating series for context-free languages it is possible to construct the structure representing all possible rule applications for any configuration.

We give a hardware design implementing some concrete examples and present the obtained experimental results which feature an important speed-up.

Host: Ion Petre

Where academic tradition meets the exciting future