Evaluation of atomic level mean force potentials via inverse folding and inverse refinement of protein structures: atomic burial position and pairwise non-bonded interactions

Two atomic level knowledge-based mean force interaction potentials(KBPs), a centrosymmetric burial position term and a long-rangepairwise term, were developed. These were tested by comparingmultiple configurations of three structurally unrelated proteinsand were found successfully to (i) discriminate native stateproteins from grossly misfolded structures in inverse foldingtests, (ii) rank identify, using the KBP energy/r.m.s.d. correlation,native from progressively less native-like (compact and dilated)structures generated via molecular dynamics sampling, providingan energy gradient sloping from partially unfolded structurestowards near-native states in inverse refinement tests, (iii)smooth the overall potential energy surface in the region ofdilated non-native structures by countering local minima ofthe in vacuo molecular mechanical potential and (iv) serve asa local minimum detector during simulated temperature quenchingstudies. These atomic KBPs discriminated native from non-nativestructures with greater overall sensitivity than did eithera residue-based pairwise interaction potential or an effectivesolvation potential based on atomic contact volume occupancy.The KBPs presented here are immediately useful as a tool forselecting ‘good refinement candidates’ from an arbitrarycollection of protein configurations and may play a role indynamic computational protein refinement.