Hi,
I have calculated the energy and the verifyScore for the conformation that has been refined with the loop refinement and sidechain refinement protocol, however, the results confused me(the modeling, loop and sidechain refinement have been conducted sequentially):

Parameters after modeling
Potential Energy = 3344142.1497 kcal/mol
Van der Waals Energy = 3354616.7206 kcal/mol
Electrostatic Energy = -15277.0148 kcal/mol
VerifyScore=177.82


Prameters after loop refinement

Potential Energy = 0.39805E+66 kcal/mol
Van der Waals Energy = 0.39805E+66 kcal/mol
Electrostatic Energy = 0.39697E+07 kcal/mol
VerifyScore=172.82

after side-chain refinement
Potential Energy = 52879.76924 kcal/mol
Van der Waals Energy = 1614.69954 kcal/mol
Electrostatic Energy = -16768.17590 kcal/mol
CHARMM ENERGY [KCAL/MOL]: INITIAL = 3.907162E+65 OUTPUT = -1416.64
VeryfyScore=155.76

Parameters for the template
Potential Energy = -6145.44795 kcal/mol
Van der Waals Energy = 5141.04743 kcal/mol
Electrostatic Energy = -16606.50818 kcal/mol

As can be seen from the above, the energy has been greatly increased after the loop refinement, while become lower after side-chain refinement. However, the VerifyScore is near the expected high score for the conformation right after modelling while become very low after side-chain refinement. Moreover, the potential energy for the template is negative, while the potential energy for the homology models is positive. But the CHARMM ENERGY seems lowered much after side-chain refinement, from the INITIAL = 3.907162E+65 to OUTPUT = -1416.64.

Which should I depend on ? or is there any other judgement for the quality of these conformations?

Thanks very much!