Recognizing misfolded and distorted protein structures by the assumption-based similarity score

A new similarity score (Sigma-score) is proposed which is able to find the correct protein structure among the very close alternatives and to distingu ish between correct and deliberately misfolded structures, This score is ba sed on and it favors the general principle 'similar likes similar', hydroph obic and hydrophilic contacts, and disfavors hydrophobic-to-hydrophilic con tacts in proteins. The values of Sigma-scores calculated for the high-resol ution protein structures from the representative set are compared with thos e of alternatives: (i) very close alternatives which are only slightly dist orted by conformational energy minimization in vacuo; (ii) alternatives wit h subsequently growing distortions, generated by molecular dynamics simulat ions in vacuo; (iii) structures derived by molecular dynamics simulation in solvent at 300 K; (iv) deliberately misfolded protein models. In nearly al l tested cases the similarity score can successfully distinguish between ex perimental structure and its alternatives, even if the root mean square dis placement of all heavy atoms is less than 1 Angstrom. The confidence interv al of the similarity score was estimated using the high-resolution X-ray st ructures of domain pairs related by non-crystallographic symmetry. The simi larity score can be used for the evaluation of the general quality of the p rotein models, choosing the correct structures among the very close alterna tives, characterization of models simulating folding/unfolding, etc.