openMM simulation notes

Recommendations from Anton for avoiding exploding polymers

Initial recommendations

  • integrator choice:
    • variableLangevin integrator recommended
    • takes an error tolerance instead of a timestep variable (e.g. 10^-4)
  • thermostat = high values high friction
  • mass_scale = argument of init of simulation (see openmmlib.py)
  • attractionRadius (be careful about interaction over other polymers).
  • wiggle_dist = higher more oscillation, less chain crossing.
  • localEnergyMinimization(maxIterations=1000,tolerance=1)
    • local energy minimization with attractive potentials might lead to super-position.
    • Try global energy minimization.

conlen = 1 nm * length_scale (=1).

potentials

  • smooth square well force
  • polynomial force
  • sticky particles should be extra hard.
  • every particle has a charge (sticky yes no) repulsion strength yes no.
  • extra repulsive force
  • try hard-core forces (very large potential energy at overlap).

dynamic adjustments (changing parameters during simulation)

  • things of interest to change: friction
  • crude way, use output as input
  • system block
  • try energyMinimization in place of localEnergyMinimization
  • avoid any collapsed pairs (they just get worse). — this is time to quit
This entry was posted in Chromatin, Software Development and tagged . Bookmark the permalink.