- what for one needs to estimate energies?
- how accurate one has to be?
- is it possible to compute energies "ab initio"?
- is it possible to extract energies from the statistics of various protein structure?
The last question will be discussed in some details. Various models of heteropolymers will be considered for comparison. It will be shown, that under some circumstances the statistics of protein structures obeys Boltzmann distribution with some effective temperature which reflects, perhaps, the evolution process of "selection" and "preparation" of the modern ensemble of protein sequences.