What I do

I realize that when I tell other people what I am doing in graduate school, they don't understand half of the words that I say. This lack of understanding can lead others (primarily my mother :) ) to describe me in terms that, while glowing, are not wholly accurate. Hence, my objective here is to clarify what exactly it is I do.

I am a computational scientist. In other words, I use computers to investigate processes that cannot be observed with traditional experimental methods. They are typically very fast, very small, and/or very slow. My current process of interest is the melting and rehybridization of small oligomers of DNA. This nanoscopic process is too fast to be observed experimentally, occurring in microseconds. Yet it is too slow to be observed using traditional methods of computer simulation where Newton's equations of motion (F=m*a) are used to evolve the system in time. It would take me dozens of years to perform one simulation that potentially could show the melting (or dehybridization) of DNA.

However, all is not lost. Brilliant men that have preceded me have developed techniques called replica exchange and parallel tempering. In these techniques, dozens of simulations are performed in parallel. At a set frequency, configurations between simulations as adjacent temperatures are swapped according to a certain acceptance criteria. In this way, these configurations perform what is termed a "random walk in temperature space." Doing so allows me to calculate the density of states as a function of temperature. From this density of states, I can deduce the melting temperature of the DNA oligomer. The value from simulation is compared with the experimental value. Ideally, the two temperatures are consistent.

To perform this replica exchange calculation, I use a program called GROMACS (Groningen Machine for Chemical Simulations). It is quite a finicky piece of software yet it is also very fast, thus ideally suited to the long simulations that need to be performed to acquire good estimates of the density of states. I've had to learn scripting languages like bash, TCL, and Python to automate the pre- and post-processing. In that way, I feel like I am very much becoming a computer scientist. I remember poking fun at computer scientists and now I am one of them!

No description of any simulation work is complete without pretty pictures. I include a picture of a 6-mer, GCATGC, that I am currently simulating. Once I obtain some meaningful results, I'll blog about those so that you all don't think I'm all hot air. Speaking of hot air, check out the PhD Comic here.