Since the functionality of the protein is determined by its 3D structure, it
would be very helpful to be able to predict the protein's structure, thus helping
to understand its role and responsibilities in the cell.
There are several approaches
towards this problem. The first, called homology, uses a protein database
to search for similar sequences of proteins. If such a protein is found (a protein
with 
of the amino-acids in its sequence being identical to the original
is usually similar enough), it is quite safe to assume that the two proteins
will have the same structure. The second approach, threading, classifies
known structures into families with similar foldings. Given a sequence of amino
acids, we select the family to which the given sequence is most likely to fold.
A third approach, de novo (from scratch), tries to find the structure
which minimizes the energy of the protein.