Given a structural description and atomic coordinates: `What does it look like?'
This is an obvious question to ask, but the value of sitting down and staring at a structure is difficult to overestimate. Of course, the pretty pictures generated by graphics packages are only representations of models, even the pretty pictures of `experimental' structures. CHARMM graphics has broad capabilities but is less convenient than some commercial packages, such as QUANTA or SYBYL. It's hard to beat RASMOL, a public domain program that spins proteins on command, with options to display them as ribbons, balls and sticks, or space-filling overlapping spheres. RASMOL also allows point-and-click atom identification as well as limited zooming and z-clipping.
Given a structural description, atomic coordinates, and an energy function: `How does the system relax and fluctuate?'
Now we're getting to the point! Structure determination is clearly a critical step toward understanding biological function, but protein function requires motion. Molecular dynamics is the link between structure and function.
We might, for example, wish to characterize the dependence of a protein's structure and dynamics on environmental conditions. We could perform simulations at different temperatures, different pressures, or different levels of hydration. We could approximate the solution environment by a periodically repeating system in which the repeating unit was a single protein in a box of water. Or the crystalline phase could be simulated as a special case of the periodic system with a particular box size and shape.
To answer questions like these on a computer, we need to employ a few techniques that manipulate the structure, , given the potential energy, .