Next week marks the beginning of my fourth and final year as an undergraduate studying biochemistry. Last year, for my bachelor’s thesis research project, I had an excellent experience working collaboratively between the labs of Martin Welch and Richard Farndale. So this year my project also involves research done in two labs: those of Jasmin Fisher, who works for Microsoft and is also a group leader in the Department of Biochemistry, and Gerard Evan, who is the head of the department (the images are copied from their websites).
Roughly, the project will entail using a freely available systems biology tool called the BioModel Analyzer (BMA; Benque et al. (2012)) to model the complex network of the (in)famous proto-oncogene Myc. The model, which I do not have to build up from scratch, is based on experimental observations about Myc. And if I understand correctly, then the next steps will involve experimenting in silico (on the computer) and then testing any new and interesting predictions that the BMA outputs in vitro.
Where does so-called executable cell biology fit in? According to Fisher & Henzinger (2007) executable biology is the “construction of computational models of biological systems”. I never thought I would work on a computer science/systems biology project and therefore a lot of the terminology and concepts are new to me (too). The first thing I learnt from reading the review is the difference between a mathematical model, which relies on (quantitative) equations, and a computational model, which relies on (qualitative) algorithms. The BMA relies on such algorithms (called qualitative networks; Schaub et al. (2007)), but has the advantage that the end user does not have to program, and that’s great because my knowledge of programming does not extend beyond the first few lessons of Python at Codecademy. Instead of coding I can drag and drop cells and proteins and connect them with activatory or inhibitory arrows to create a network. The program subsequently analyses whether the system stabilises/reaches a steady state in which the concentration values of all variables (i.e. proteins) reach a fixed value. And if I understood correctly then the interpretation is that a stable model is biologically feasible, whereas a model that does not stabilise cannot be expected to occur in nature [EDIT I have been corrected on this: unstable models can occur in nature, for example in the form of an oscillation or in a case where there is more than one stable outcome]. So much for an introduction to the project; once I am less confused myself there will be an update.
Benque D, Bourton S, Cockerton C, Cook B, Fisher J, Ishtiaq S, Piterman N, Taylor A, Vardi MY. (2012) BMA: Visual tool for modeling and analyzing biological networks. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 7358 LNCS, pp. 686-692.
Fisher J, Henzinger TA (2007) Executable cell biology. Nature Biotechnology 25: 1239-1249
Schaub MA, Henzinger TA, Fisher J (2007) Qualitative networks: a symbolic approach to analyze biological signaling networks. Bmc Systems Biology 1