The convergence-adaptive enhanced sampling method

The convergence-adaptive roundtrip enhanced sampling method enables rapid and accurate FEP calculations.

Procedures to run simulaitons of a thermodynamic process with the CAR method are as follow:

1. Step 1: Prepare the input files.
   • coordinate file and topology file;
   • a json file specifies the lambda scheme; e.x. input_temp/lambdas.json
   • a json file includes input settings for AMBER; e.x. input_temp/input.json*. Note: SCMASK should be modified to unique atoms list for each pair as input.json shown in directory of each pair.
   • settings of CAR method; e.x. input_temp/car_run_input.txt. Note: Some settings need to be changed according to the working directory path and filenames.
2. Step 2: Start the CAR enhanced adaptive sampling program.
   • python $CAR_SCRIPT_DIR/car_converge_control.py -i car_run_input.txt

Details of car_run_input.txt

## input file for convergence adaptive control md run
[normal_alc_md]
simulation_software     = amber # the simulation software
coordinate_file         = __COOR__ # the initial input coordinate_file, can be '.prmcrd' or '.rst7'
topology_file           = __TOPO__ # the initial input topology_file
prod_md_time            = 10 # simulation time of each state in unit of ps
mbar_lambda_dict_file   = lambdas.json # the lambda schemes
input_file              = input.json # settings for simulations

[segmented_md_control]
segment_lambda_step     = 0.2 # determine how to divide the thermodynamic process into parts
num_neighbors_state     = 5 # the number of neighboring states to record potential energies when simulating specified state.
min_reitera_times       = 2 # specify the minimun reiteration times (an iteration includes a switchback round and a restart forward round) of a part
max_reitera_times       = 50 # specify the maximum reiteration times (an iteration includes a switchback round and a restart forward round) to terminate simulations of a part
error_max_lambda_0to1   = 0.15 # specify the uncertainty allowed for an entire thermodynamic process
ifrun_preliminary_md    = True # whether to run pre-equilibrium MD run at initial state 0
ifuse_initial_rst       = True # whether to start MD run with initial input coordinate_file, else specify a file_path to run continuous MD simulation
ifoverwrite             = True # if True, history directories will be removed, else concat the results of new simulation to existed files.

Data available

The input and structure files of benchmark for the CAR enhanced adaptive sampling can be obtained at scidb.cn.