Protein structure prediction
One of the major tasks in bioinformatics is the prediction of protein structures. In this case we have several DNA sequences (primary structure) and would like to know the secondary / third level structure for it. How can I do that?
After a first glimpse, reading the
Wikipedia article for it, I got the impression this field divides into two major groups:
Ab initio protein modelling
The group of methods which "just" uses physics and calculates the interaction between each atom to predict the whole structure of the molecule. This uses a vast amount of computational power, which is the reason for projects like Blue Gene or folding@home.
Comparative protein modelling
The group of methods which compares unknown protein sequences to protein sequences with known structure, assuming a similar behavior in folding. There a millions of different proteins but there seems to be a limited number of ways a protein can fold to. In other words the set of different tertiary structural motifs is limited.
Also one interesting method to get higher protein structures from known proteins is to use protein-protein docking algorithms.
There seems to be a bunch of software to do protein prediction. There is also a test for comparing these approaches and algorithms which is called CASP and takes place every two years. Looks like fun what these Russian guys at the prediction center do: Maths :-)
