Background

Proteins are long chains of Amino Acids (AA). There are 20 types of AA that serve as building blocks for proteins. Each AA has a specific chemical structure. The length of a protein chain can range from 50 to 1000-3000 AA (200 on the average). One of the interesting properties of proteins is the unique folding. The AA composition of a protein will usually uniquely determine (on specific environment conditions) the 3D structure of the protein (e.g., two proteins with the same AA sequence will have the same 3D structure in natural conditions). Researches of 3D structure of proteins have shown that when a folded protein is artificially stretched to a chain, it folds back to it's original 3D structure. Proteins are known to have many important functions in the cell, such as enzymatic activity, storage and transport of material, signal transduction, antibodies and more. All proteins whose structure is known are stored in the

Protein DataBank (PDB) which contains about 10,000 proteins [6].

A protein has multiple levels of structure (see figure13.1):

• Primary structure - Chain of Amino Acids (1 dimensional)
• Secondary structure - Chains of structural elements, most important of which are $\alpha$-helices and $\beta$-sheets.
• Tertiary and Quaternary structure - 3D structure, of a single AA chain or several chains, respectively.

  

Figure 13.1: The four structuring levels of a protein. Taken from http://gened.emc.maricopa.edu/bio/bio181/BIOBK/BioBookCHEM2.html

The common methods for finding protein 3D structure are:

• X-ray diffraction and neutron-diffraction
• Nuclear Magnetic Resonance

These methods are slow and costly (taking up to several months of lab work), much slower than DNA sequencing. This creates interest in algorithms for protein structure prediction.