Protein folding dynamics: Designing transferable interaction potentials and developing New monte carlo and molecular dynamic protein models with different level of complexities

Protein folding is one of the most challenging problems in biological sciences. Depending on amino acid contents, chain length, native topology, etc., proteins exhibit experimentally different thermodynamic and kinetic behaviors such as the level of cooperativity and diversity in the folding rates. Understanding of these phenomenon requires detailed analysis of theoretical protein models with different level of complexities. In the first part of this thesis, the effects of intra-chain interaction networks on protein folding thermodynamics and kinetics are investigated by coarse-grained molecular dynamic models for 13 different proteins. It was found that the level of thermodynamic cooperativity from downhill to two-state behavior remarkably depends on the average connectivity of the intra-chain interaction network. A newly defined contact order parameter which takes into account angular interaction terms besides the non-bonded interactions (COWAT) clarifies both the kinetic and thermodynamic behaviours. At the second part, new transferable interaction potential is developed for all-atom Monte Carlo protein models. From data mining 25,801 protein folded structures, statistical potential for: (i) two-body interactions, (ii) NHO hydrogen bonds, (iii) 7 different local consecutive residue triplets (iv) local consecutive residue quadruples terms are generated. The effect of the chain distance between two atoms on their two-body interaction is further investigated and total unique pair size is increased to 195,223. Two studied alpha-structured protein is modeled with Monte Carlo simulation are both successfully folded into their native structure, with a lowest RMDS value 2.29A and cooperative folding-unfolding transition is observed. Keywords: Protein Folding, Protein Intra-chain Interaction Networks, Molecular Dynamics Go Model, Transferable protein models, All-Atom Monte-Carlo Protein Models