Computational and mathematical biology
Colin Johnson, University of Kent at CanterburyOne of my main areas of research is the application of computational, mathematical and statistical techniques to understand biological and medical problems.
My main interest in theoretical biology is in using computational and mathematical methods to understand signal transduction. This is the set of processes in organisms by which different components of the organism communicate by sending chemical messengers. Such signals have many roles within the body, such as encouraging growth during development, controlling repair of cells following injury, detection of infection, controlling the link between food intake and hunger, et cetera. This system can go wrong in many ways; most seriously problems in the growth signalling system can lead to the development of some form of cancer.
In collaboration with colleagues in the biosciences department here at Kent and at Cambridge, we have been developing models of these signal transduction processes. In particular we are interested in understanding how the size and structure of the receptors on the cell surface which receive signals influence signal strength and the "gain" of the signal. Current work focuses on matching such simulations to real exeperimental data. Using image analysis techniques we are able to get time series of receptor cluster sizes and composition in real cells, and we are applying optimization techniques to match the simulation to these data. This will enable us to estimate parameters in the real system that are inaccessible to direct measurement, such as probabilities of protein-protein binding, and enable us to carry out an analysis of the stability of the system in order to identify components of the system that are most suited to therapeutic intervention.
Such modelling of signalling can also be used in biologically inspired computing. My research student Miguel Mendao has been working on robot control architectures that combine neural networks with a simulation of the endocirne system, to facilitate task selection. Another project is using signalling systems as the programming language in a genetic programming-style system.
Reconstructing Systems from FactsA number of current projects are concerned with constructing models of biological systems given certain facts. In a project with Alex Freitas and our research student Mudassar Iqbal, we are looking at reconstruction of protein interaction networks for e.g. microarray data. Another project is concerned with describing qualitative facts about a system of biochemical interactions as temporal logic statements, and then using techniques such as model checking to analyse whether proposed models (in the other sense!) of the system are consistent with these facts.
Non-classical Computing and Emergence
An more abstract interest is in the concept of non-classical computing. This is concerned with regrounding computing by considering the computational properties of real-world objects, rather than grounding the idea of computing in notions of abstract computing machines. By doing this we want to (i) be able to produce new kinds of computers which exploit other aspects of physics and chemistry than traditional computing and (ii) understand some aspects of the world (e.g. medical problems) by considering them as problems about the computational capabilites of those systems. This area has recently been given a boost by the UK Computing Research Committee "grand challenge" project in this area.
Another related interest is in the idea of emergent phenomena, in particular understanding what kind of statement is being made when we say that something is emergent. By comparison with approaches to concepts which have similar ontological difficulties (for example the idea of vagueness in language, and the Sorites paradox) we are attempting to get a better understanding of the status of the concept of emergence.
Other ProjectsOver the last few years I have carried out a number of other projects. In collaboration with Jacqueline Whalley from Auckland and Mick Tuite from UKC, we have been applying simulation techniques to understand the development of prion proteins. Other projects have been concerned with novel approaches to bioinformatics data analysis, for example applying color-based visualisation techniques to spot common patterns in protein sequences, applying ant-colony optimization techniques to gene sequence analysis and the development of novel bio-inspired data mining techniques for the analysis of proteins. I also occasionally do more straightforward statistical analysis and data mining, for example I have recently been working with psychologists at Liverpool on a geometrical approach to facet theory for multivariate data analysis and work on classification in structural biology.
Colin Johnson University of Kent at Canterbury