Scalable Software Services for Life Science

DISCRETE Use Case Scenarios

DISCRETE is a package devised to simulate the dynamics of proteins using the Discrete Molecular Dynamics (DMD) methos. The difference with respect to a standard molecular dynamics simulation is that the particles are considered to move with constant velocity until a collision occurs. Upon each collision there is a transfer of linear momentum between the colliding particles, where total momentum and energy are conserved. As the particles move in ballistic regime, the equations of motion do not have to be integrated. This leads to a drastic saving in computing time. In DMD the interaction potentials are stepwise potentials, and the events (collisions) occur when the distance between two particles corresponds to a step in its interaction potential. DMD is an event-driven method, so the timestep is not predefined like in a standard molecular dynamics simulation, but it is the time between consecutive collisions. The frequency of the collisions in the system increases linearly with the number of particles, and with the number of steps in the interaction potential.

Alternative applications

We have developed variations of DMD, the molecular dynamics code of DISCRETE, devised to deal with conformational transitions proteins (DMDIMS) and protein-protein docking (DMDOCKING)

Conformational transitions of proteins

DMDIMS generates trajectories for the conformational transition between two structures of the protein that are given as input files. The difference with the standard DMD is a Metropolis algorithm where the code analyzes if the simulation is moving towards the target conformation. Therefore the simulation is the addition of short DMD trajectories that are Monte Carlo steps that have been accepted. When one of these trajectories is rejected by the Metropolis algorithm, the trajectory is rewind to the coordinates at the beginning of the Monte Carlo step, but with the current velocities, so the simulation does not enter in a closed loop. The main advantage of this method respect to other methods to generate conformational transitions is that the potential energy remains always low, since no external potential has been applied to the system to enforce the transition.

Protein-protein docking

DMDOCKING is a version of DMD devised to refine protein-protein docking poses. The program uses a multi-scale  description of the proteins: all-atom resolution at the interface of the proteins, and a residue-level resolution in the rest of the protein (only alpha carbons).  This leads to a drastic decrease in the number of particles included in the simulation, therefore the simulation is accelerated.
Structure-based potentials are defined in the alpha-carbon represented region of the proteins in order to conserve the structure. Since this is a code to refine rigid docking poses, constraints are defined between the two proteins to hold the relative positions of the proteins.  All-atom resolution is necessary only at the interface because the scoring function on protein-protein docking is computed from the interprotein interactions. The relevant output of DMDOCKING are the different energy terms due to the different types of interaction between the atoms of each protein: solvation, electrostatic and Van der Waals. The scoring function used to rank the different poses is a linear combination of these terms.