Abstract
It is well known that sequences of numerous proteins fold into essentially the same structure. This observation is the reason behind the success of homology in modeling protein structures. However, it is not clear how "many" sequences is "many"? The question I will address is the accurate computations of the number of sequences that match a given fold (accuracy is bound by the quality of the energy function). I will present an algorithm that makes exact counting in sequence space possible. An application to non-degenerate set of 3400 structures from the Protein Data Bank will be described in details. Deviations from expected exponential behavior are discussed, and the calculation of the temperature of evolution is illustrated.