Between 2002-2006 I was studying towards a PhD in Bioinformatics/Computational Biology at University College London with CoMPLEX (Centre for Mathematics and Physics in the Life Sciences and EXperimental Biology) and the Institute of Child Health. My supervisors were Dr. Mike Hubank, Prof. Jaroslov Stark and Prof. Robin Callard. I was based on a project that was using microarray data to construct gene regulatory networks. The specific network I am looking at is the p53 DNA damage network, this is an important network that controls whether a cell should die or not after stress. In about 50% of cancer types the system breaks down. More information about the p53 network can be found here.

In October 2002 I completed an MRes in biological complexity with CoMPLEX. That year involved learning biology from scratch and also mathematical modelling techniques. As part of the course I produced three essays/miniprojects and a summer project.

My thesis is titled "Modelling the p53 Gene Regulatory Network" and covers three main areas of research. In the first part of the thesis ordinary differential equations models of protein expression in the system are proposed based on previous biological knowledge. In the p53 system the regulation of the location of the core components appears to be important. Therefore, a model is proposed that takes into account some of these localisation mechanisms. This model is examined to determine whether the inclusion of localisation regulation improves the performance of the system.

The second part describes studies into parameter estimation using a small experimental protein data set. To make predictions based on these models it is important to determine how well the behaviour of the model matches biological data, i.e. to find the parameters that cause the model solutions to best fit the data. This is important as it is rare to have direct parameter measurements in this kind of system. Established techniques such as simulated annealing are implemented and examined along with a novel technique based on linear algebra, collocation and a series of B-splines. A number of the proposed models are evaluated based on the latter technique. It is easier to gather data on the amount of mRNA than the amount of functionally active protein.

In the final part of the thesis, a simple model of positively regulated gene expression is investigated with a view to gaining information about the protein level response to DNA damage using microarray mRNA data. For each gene, a quantity called the G time profile is proposed which is representative of the time profile of the activity that is driving the quantity of that gene's mRNA. This quantity can be used to group genes that share the same transcription factor and find the main "activities" that drive the DNA damage response.

I was also worked as a marker and demonstrator for the mathematics department.