What are the similarities and differences between "template based" and "template free" methods for protein structure prediction?

In fact, the key difference between "template based" and "template free" structural prediction methods lies in the different conformational initialization methods. Template based prediction method obtains the initial conformation by searching and recognizing the known structure which has homology or structural similarity with the target protein as the template; template free prediction method usually constructs the initial conformation from the beginning with a small structure fragment.

At present, no matter which method is used, the known structural information in PDB database will be used to construct an initial conformation which is far better than random conformation, thus significantly reducing the pressure of conformation search, including reducing the amount of calculation and reducing the dependence on high-precision force field

Explanation:

http://www.creative-biostructure.com/protein-crystallization-and-structure-determination_13.htm