Automated Scripts in Computational Biology
The complexity built in to a biological system makes it hard to simplify and it is when a very simplified mathematical model reflects the actual experimental behavior (+/- the noise), mathematicians and physicists get very excited - they enjoy sharing and describing the model. A trained experimentalist/biologist will have the intuition to drive a simplified model.
A computational biologist, depending on his interests, tries to simplify a complex system or adds complexity to a simple model - sometimes, he does both. The behavior of a chosen biological systems may be simple / not - we can only know that when we start measuring the system’s output with respect to the various parameters that we think may affect the underlying biology. Often times, that means, simulating the computational model in a wild number of combinations of parameter values.
For example, I may be interested in how much an addition of 0.01 units to a parameter p1 affects the output response of the system. After seeing the output, I may get another idea/hypotheses and ask how much variations in p2 may diminish the changes effected by p1 to the output response. In this way, the final combinatorial space we explore gets very high.
Computationally speaking, that would mean doing a lot of ‘mkdir’s, chmod’ing, ‘cp’ n ‘mv’ing, ‘emacs’ng, reorganizing —- all this for adding/subtracting that 0.01 units to some tiny variables. This unfortunately, consumes a lot of human attention and over time, very tiring. In the end, we may sometimes find ourselves in a situation where, if we wonder for a second about what brought us to the current idea that is driving our analysis, we may not know the answer. At that point, we are desperately in need of some automated scripts or if its not out there, build it up ourselves.
For someone in computer science, this may not be a huge challenge. I just wanted to suggest to any biologist who is out there trying to do such phase space analysis to use Linux or Windows+Cygwin(XWinServer) and the several simple tools they come with and consult websites like StackOverflow. Particularly, the following linux tools/commands are very useful for a lot of these tasks: Sed, Awk, grep, cut and other similar functions. Combine these with the computing power of Matlab and you can build a suit by yourselves. Matlab seamlessly integrates with Linux command line via the ‘system’ command.