Local conformational space

A force field can be coupled with molecular dynamics (MD) algorithms to sample local conformational space around the energy minimum of a macro-molecule. An MD algorithm involves the numerical integration of Newton s equation s of motion, where the acceleration of a particle is a = Fj/w, and the force Fj is the negative gradient of the potential surface. New atomic positions and velocities are calculated repeatedly with a small time step on the order of 10 " sec (1 fsec). The resulting set of time dependent coordinates is known as a trajectory. 

A different technique used for exploring conformational space is the molecular dynamics method. While it is effective for searching local conformational space, it requires remarkably more computer time than the other methods for global search problems. ... 

For a conformation in a relatively deep local minimum, a room temperature molecular dynamics simulation may not overcome the barrier and search other regions of conformational space in reasonable computing time. To overcome barriers, many conformational searches use elevated temperatures (600-1200 K) at constant energy. To search conformational space adequately, run simulations of 0.5-1.0 ps each at high temperature and save the molecular structures after each simulation. Alternatively, take a snapshot of a simulation at about one picosecond intervals to store the structure. Run a geometry optimization on each structure and compare structures to determine unique low-energy conformations. 

This algorithm will generate the ensemble of conformations at thermodynamic equilibrium, if the conformational variations introduced in step 1 (the so-called moves ) are sufficient to cover all possible regions of conformational space and are locally reversible. It is easy to understand that feature of the Metropolis procedure the transition probability into the higher energy state is given by the Boltzmann factor, the transition probability into the lower energy state is one. The relative population of the two states is then the ratio of the transition probabilities, thus the Boltzmann factor this is the outcome expected from thermodynamics. 

In contrast to measures of convergence that reflect a single local observable, for example, a torsion angle, some methods focus on the global sampling quality. For a simulation of a macromolecule, the distinction would be between asking "how well do I know this particular quantity " and "how well have I explored the conformational space of the molecule " The latter question is critical, in that if the conformational space is well sampled, most physical quantities should be known well. 

---