Being able to predict the tertiary structure of a protein from its primary sequence is one of the most important goals for protein researchers, but has, so far, been a task beyond capabilities. But a study by Simmerling, Strockbine and Roitberg in The Journal of the American Chemical Society shows for the first time how this could now be possible using computer simulations.

The researchers looked at the smallest protein known to show normal folding properties — a 20-residue sequence called 'Trpcage', created by Niels Andersen and colleagues — and carried out a series of molecular dynamics simulations, starting only from the primary sequence of Trpcage. The best structure was chosen purely on the basis of lowest energy, not from any structural features or sequence similarity to known protein structures. This was then sent back to the Andersen group before the NMR structural data was released.

The correlation between the two sets of data was impressive. Structural details, including the arrangement of the hydrophobic core and a short α-helix, were picked up by the simulation. The correspondence between the chemical shift deviations was also good, and both methods showed that the charged terminal residues adopted multiple conformations. The computer simulation also uncovered structural details that were not apparent from the NMR data, such as an Asp–Arg bridge, the presence of which is supported by other experiments.

So, it seems that computer simulations could at last be used to predict the three-dimensional structure of a protein accurately. Also, the use of energy as a successful scoring function suggests that we understand the underlying interactions that determine folding. And although there are obvious challenges when applying this to larger, more complex proteins, this could be a very important tool for drug design.