The problem of how a 1-D poly peptide chain acquires the form of a native 3-D protein structure is referred to as the protein folding problem.
This is one of the most challenging and important research areas in Biochemistry. All the information needed to specify the protein's 3-D structure is contained within its amino acid sequence, and given suitable conditions, most proteins will spontaneously fold into their native states.
Although studies have provided much information on the folding process, our understanding of a way proteins fold is not full yet.

1. The main reasons the protein folding problem is considered such a difficult one are:if a protein had to search all the possible conformations to find the energetically most favorable one, even a small protein would have to search for an unreasonable amount of time. Since it is known that proteins usually fold into their native states within much shorter time, it is clear that the protein retains partially correct intermediate rather than randomally scan all its possible conformations. Therefore, the energy calculations involve millions of atomic interactions, and going over all of them to find the most favorable conformation is beyond the ability of even the most powerful computers available today.
2. Structure is much more conserved than sequence. Similar sequences usually give similar structures, but different sequences may lead to a similar structures. As a result, there is a much smaller number of known structures than sequences. This means that the relationship between sequence and structure is not one-to-one, and the tules that predict the structure out of the sequence are not fully understood.

• Experimental methods:
• NMR (nuclear magnetic resonance)
• X-ray Crystallography
• Theoretical and computational methods:
• Homology modeling assumes that similar sequence leads to similar structure. It involves fitting a sequence to a known 3-D structure of a similar sequence. Two proteins are considered homologous if they have identical amino acid residues in a significant number of sequential positions along the polypeptide chains (usually at least 30-40% of identity). The results of homology modeling are more satisfying than those of ab-initio modeling methods. About 2% of the known structures were determined this way.
• Ab initio methods do not rely on known structures, rather they use the physics of interactions among atoms, including the solvent, to define a force field and then use methods like molecular dynamics or monte carlo to determine the most stable structure.
• threading (inverse folding): Since it is assumed that there is only a limited number of stable folds and many different sequences having the same fold, the inverse folding method tries to figure which specific patterns are compatible with a specific fold. The used methodology is called threading, since it involves "threading" a specific sequence through all known folds and estimate the probability that the sequence can have such a fold.