Thursday, May 29, 2008

Physics rules the world

One of my college professor used to tell us that physics was the MOST important science to understand life because all the other ones rely on physical concepts. I did not like physics too much and decided to pursue a career involving biology. Soon enough I realized that physics was indeed at the core of almost everything. From the aqueous solutions to the cell membrane physics define what really happens.

Protein folding is no exception. Physical forces influence the interaction of thousands of atoms and shape polypeptides. No function! The problem with a physical approach to understand and model protein folding resides in the fact that femtosecond calculations have to be made. As Ken Dill (UCSF) puts it, it is all about timing:

"You need to be faithful to the atomic dynamics from one instant to the next, otherwise you violate Newton's laws"

Kenneth Dill is part of a group that participated in a protein-folding competition: CASP7 (Critical Assessment of Techniques for Protein Structure Prediction) held in november 2006 in California (Pacific Grove). Basically, each team (there were 250 teams) had to predict the structure of 6 three months! To do so, many, including Dill, used supercomputers such as Tungsten at NCSA. To predict protein folding, Dill and his group used the ZAM model (Zip And Assembly), once a few amino acids find their position, the rest of the amino acids in the polypeptide interact with each other in a Zipper fashion.

Fig 1: ZAM predictions in CASP7 compared with experimental PDB structures. The GDT (CASP's Global Distance Test) gives the percentage of residues (x-axis) whose alpha carbon coordinates lie within a given cutoff distance (y-axis) from the native structure, for predictions by all participants in CASP7 (orange colors). The best predictions correspond to lines in the lower-right quadrant of the graph. The five ZAM models are shown for each target in gray, with the best model highlighted in red. (click here to access original Website)

The physical approach allowed adequate prediction for 4 of the 6 proteins (see figure 1) and Dill thinks his group did at least as good, if not better, that other groups using bioinformatics.

Yes...I have to admit, physics still has a very bright future!

For a list of NCSA projects click HERE

Image source: