Framework for loop structure prediction.
The most important classes in this namespace are LoopProtocol and LoopMover. LoopMover provides an interface for loop sampling algorithms. Methods are provided for specifying which loops need to be sampled and for setting up a compatible fold tree. Some important LoopMover subclasses include RepackingRefiner, MinimizationRefiner, and KicMover. The LoopProtocol class provides a framework for combining any number of loop movers into a single Monte Carlo simulation. Aspects of the simulation like the number of iterations, the score function, and the temperature can be easily controlled.
Below is an excerpt from protocols::comparative_modeling::LoopRelaxMover that shows some of the classes from this namespace in action. This code creates a protocol featuring KIC moves followed by a rotamer trials and local minimization. Full repacking is done once every 20 iterations. The number of iterations can be controlled from the command line, and a concise progress update will be reported after every iteration.
if (option[OptionKeys::loops::repack_period].
user()) {
}
protocol->add_mover(
new PeriodicTask(
new RepackingRefiner, repack_period));
protocol->add_mover(
new RotamerTrialsRefiner);
protocol->add_mover(
new FullatomMinimizationRefiner);
if (option[OptionKeys::loops::outer_cycles].
user()) {
outer_cycles =
option[ OptionKeys::loops::outer_cycles ]();
}
if (option[OptionKeys::loops::max_inner_cycles].
user()) {
inner_cycles = std::max(inner_cycles, max_cycles);
}
if (option[OptionKeys::loops::fast]) {
outer_cycles = 3;
inner_cycles = 12;
}
protocol->set_score_function(fa_scorefxn_);
protocol->set_iterations(outer_cycles, inner_cycles, 2);
1.8.7