The objective of the proposed research is to develop a new computational technique LES (Locally Enhanced Sampling) for computer simulation of biomolecules. In the new approach a large number of parallel trajectories are run for a small part of principal interest for which extensive sampling is required (e.g. for a diffusing ligand or a fluctuating side chain). At the same time a single trajectory is evaluated self-consistently for the rest of the system (e.g. the protein and the surrounding solvent). LES reduces the computational effort compared to the usual molecular dynamics by one to two orders of magnitude, while maintaining the detailed atomic level description of the system. It thus enables detailed theoretical studies of biological systems whose investigation has been limited until now by the prohibitive computational cost. The research will focus first on methodological problems. Several extensions of the technique are proposed, including application of LES as a tool for efficient thermodynamic averaging. LES will be then employed in a systematic search for ligand diffusion paths in the proteins of the globin series (myoglobin, leghemoglobin, horse and human hemoglobin, erythrocruorin). Due to the computational efficiency of the method it will enable to examine evolutionary effects on the dynamics of a biochemical process for the first time. The second application is to rates of side chain transitions in proteins including the arginine 45 transition in myoglobin. The possible role of the latter process in the biological activity of myoglobin will be examined.