From 60ef1111c6f0628993ce20963b771c7ba0eb7616 Mon Sep 17 00:00:00 2001
From: cryobry <38270216+cryobry@users.noreply.github.com>
Date: Sat, 6 Jul 2019 21:13:14 -0400
Subject: [PATCH] Initial commit

---
 .gitignore       |   4 +
 functions.py     |  28 +++++
 job.py           | 152 +++++++++++++++++++++++++
 requirements.txt |   3 +
 run.py           |  40 +++++++
 simulate.py      | 282 +++++++++++++++++++++++++++++++++++++++++++++++
 6 files changed, 509 insertions(+)
 create mode 100644 .gitignore
 create mode 100644 functions.py
 create mode 100644 job.py
 create mode 100644 requirements.txt
 create mode 100755 run.py
 create mode 100644 simulate.py

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..1393696
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,4 @@
+auto_namd/
+__pycache__
+testing
+hosts.sh
diff --git a/functions.py b/functions.py
new file mode 100644
index 0000000..19d8c55
--- /dev/null
+++ b/functions.py
@@ -0,0 +1,28 @@
+#!/usr/bin/env python3
+import os
+import re
+import argparse
+
+
+def natural_key(string_):
+    return [int(s) if s.isdigit() else s for s in re.split(r'(\d+)', string_)]
+
+
+def abs_path(path):
+    abs_path = os.path.abspath(path)
+    return abs_path
+
+def parser():
+    parser = argparse.ArgumentParser()
+    parser.add_argument('--steps', type=int,
+                    help='Number of steps to run the simulation')
+    parser.add_argument('--jobs_path', type=str, nargs='+', action='append',
+                    help='Jobs path containing the jobs to be run')
+    parser.add_argument('--namd_params', type=str,
+                    help='Optional parameters to send to namd')
+    parser.add_argument('--ffs_path', type=str,
+                    help='Location of the forcefield files')
+    parser.add_argument('--namd2_bin', type=str,
+                    help='Location of the namd2 executable')
+    args = parser.parse_args()
+    return args
diff --git a/job.py b/job.py
new file mode 100644
index 0000000..ebe09cc
--- /dev/null
+++ b/job.py
@@ -0,0 +1,152 @@
+import os
+import sys
+import glob
+
+from functions import natural_key, abs_path
+
+class Job:
+    """
+    A class that represents a job/system residing in a unique directory path
+
+
+    Calling prepare_sim() assigns the additional instance variables:
+        steps = number of steps to run the simulation
+        ffs = a list of forcefield files for the simulation
+        next_step, final_step_num = the type of step and the final number of steps
+        conf, out file = the configuration file and the simulation output file
+    """
+
+    def __init__(self, path):
+        self.set_empty_vars()
+        self.path = path
+        self.name = self.get_job_name()
+        self.prefix = os.path.join(self.path, self.name)
+        self.pdb, self.psf = self.get_pdb_and_psf()
+        self.stage, self.step, self.coor = self.get_stage()
+
+
+    def set_empty_vars(self):
+        self.pdb, self.psf = '', ''
+        self.steps = ''
+        self.ffs_path, self.ffs = '', ''
+        self.next_stage, self.next_step = '', ''
+        self.conf, self.out = '', ''
+
+
+    def get_job_name(self):
+        '''Get job name from path'''
+        name = os.path.basename(self.path)
+        return name
+
+
+    def get_pdb_and_psf(self):
+        pdb = glob.glob(f'{self.prefix}*_solv_ion.pdb')
+        psf = glob.glob(f'{self.prefix}*_solv_ion.psf')
+        if len(pdb) == 1 and len(psf) == 1:
+            return pdb[0], psf[0]
+        pdb = glob.glob(f'{self.prefix}.pdb')
+        psf = glob.glob(f'{self.prefix}.psf')
+        if len(pdb) == 1 and len(psf) == 1:
+            return pdb[0], psf[0]
+        else:
+            print('No PDB or PSF files found, exiting...')
+            sys.exit(1)
+
+
+    def get_stage(self):
+        coors_path = self.prefix + '*.coor'
+        coors = glob.glob(coors_path)
+        coors.sort(key=natural_key)
+        if len(coors) >= 1:
+            coor = coors[-1]
+            stage = coor.split('_')[-2].split('.')[0]
+            step = int(coor.split('_')[-1].split('.')[0])
+        else:
+            coor = ''
+            stage = ''
+            step = 0
+        return stage, int(step), coor
+
+
+    def prepare_sim(self, ffs_path, steps):
+        self.steps = steps
+        self.ffs_path = abs_path(ffs_path)
+        self.ffs = self.get_ffs()
+        self.next_stage, self.next_step = self.get_next_stage()
+        self.conf, self.out = self.conf_out_file()
+
+
+    def get_next_stage(self):
+        if self.stage == '4-sim':
+            next_stage = '4-sim'
+            next_step = self.step + self.steps
+        elif self.stage == '3-heat':
+            next_stage = '4-sim'
+            next_step = steps
+        elif self.stage == '2-min':
+            next_stage = '3-heat'
+            next_step = 0
+        elif self.stage == '1-min':
+            next_stage = '2-min'
+            next_step = 0
+        elif self.stage == '':
+            next_stage = '1-min'
+            next_step = 0
+        return next_stage, int(next_step)
+
+
+    def get_ffs(self):
+        ffs = glob.glob(os.path.join(self.ffs_path, '*'))
+        return ffs
+
+
+    def conf_out_file(self):
+        if self.next_stage == '4-sim':
+            conf = f'{self.prefix}_{self.next_stage}' \
+                        f'_{str(self.next_step)}.conf'
+            out = f'{self.prefix}_{self.next_stage}' \
+                       f'_{str(self.next_step)}.out'
+        else:
+            conf = f'{self.prefix}_{self.next_stage}.conf'
+            out = f'{self.prefix}_{self.next_stage}.out'
+
+        return conf, out
+
+
+    def info(self):
+        print(f'Job Path: {self.path}\n'
+              f'Name: {self.name}\n'
+              f'Working PDB: {self.pdb}\n'
+              f'Working PSF: {self.psf}\n'
+              f'Previous Step: {self.stage}\n'
+              f'Previous Step Number: {str(self.step)}\n'
+              f'Next Stage: {self.next_stage}\n'
+              f'Next Step Number: {str(self.next_step)}\n'
+              f'FF Path: {self.ffs_path}\n'
+              f'FFs: {self.ffs}\n'
+              )
+
+
+
+def get_jobs_from_path(jobs_path):
+    jobs_path += '/**/*.pdb'
+    job_dirs = glob.glob(jobs_path, recursive=True)
+    job_dirs_l = []
+    for job_dir in job_dirs:
+        job_dirs_l.append(os.path.dirname(job_dir))
+    if len(job_dirs_l) < 1:
+        print('No valid jobs found in jobs path!')
+        sys.exit(1)
+    return job_dirs_l
+
+
+def get_next_job(jobs):
+    jobs.sort(key=lambda x: (x.stage, x.step))
+    return jobs[0]
+
+
+def create_job_instances(job_paths):
+    job_instances = []
+    for job_path in job_paths:
+        job_instances.append(Job(job_path))
+    return job_instances
diff --git a/requirements.txt b/requirements.txt
new file mode 100644
index 0000000..f9d9b6c
--- /dev/null
+++ b/requirements.txt
@@ -0,0 +1,3 @@
+git+https://github.com/Eigenstate/vmd-python.git
+numpy
+
diff --git a/run.py b/run.py
new file mode 100755
index 0000000..af84d3c
--- /dev/null
+++ b/run.py
@@ -0,0 +1,40 @@
+#!/usr/bin/env python3
+
+import glob
+import os
+
+from job import get_next_job, get_jobs_from_path, create_job_instances
+from simulate import Simulate
+from functions import abs_path, parser
+
+
+def main(jobs_path, ffs_path, steps, namdbin, params):
+
+    jobs_path = abs_path(jobs_path)
+    job_dirs_l = get_jobs_from_path(jobs_path)
+    jobs = create_job_instances(job_dirs_l)
+    job = get_next_job(jobs)
+    job.prepare_sim(ffs_path, steps)
+    job.info()
+    Simulate(job, namdbin, params)
+
+
+if __name__ == '__main__':
+
+    args = parser()
+
+    if args.jobs_path is not None and os.path.isdir(args.jobs_path):
+        jobs_path = args.jobs_path
+    else:
+        jobs_path = '/home/bryan/MD/CFTR/4-jobs'
+
+    if args.namd_params is not None:
+        params = args.namd_params
+    else:
+        params = '+p7'
+
+    main(jobs_path=jobs_path,
+         ffs_path='/home/bryan/MD/CFTR/0-forcefields',
+         steps=1000000,
+         namdbin='/home/bryan/bin/namd2',
+         params=params)
diff --git a/simulate.py b/simulate.py
new file mode 100644
index 0000000..be05b6f
--- /dev/null
+++ b/simulate.py
@@ -0,0 +1,282 @@
+import subprocess as sp
+import vmd
+
+class Simulate:
+
+    def __init__(self, job, namdbin, params):
+        self.prepare(job)
+        self.run_namd(job, namdbin, params)
+
+    def prepare(self, job):
+        if job.next_stage == '1-min':
+            self.min1(job)
+        if job.next_stage == '2-min':
+            self.min2(job)
+        if job.next_stage == '3-heat':
+            self.heat(job)
+        if job.next_stage == '4-sim':
+            self.sim(job)
+
+    def run_namd(self, job, namdbin, params):
+        cmd = f'{namdbin} {params} {job.conf} > {job.out}'
+        print(f'Running: {cmd}')
+        p1 = sp.run(cmd, shell=True)
+
+    def min1(self, job):
+        steps = 15000
+        with open(job.conf, 'w') as conf:
+            conf.write('# Input\n')
+            conf.write(f'structure {job.psf}\n')
+            conf.write(f'coordinates {job.pdb}\n')
+            conf.write('paraTypeCharmm on\n')
+            for ff in job.ffs:
+              conf.write(f'parameters {ff}\n')
+            conf.write('\n')
+            conf.write('# Temperature\n')
+            conf.write('temperature 0\n')
+            conf.write('\n')
+            conf.write('# Force-Field Parameters\n')
+            conf.write('exclude scaled1-4\n')
+            conf.write('1-4scaling 1.0\n')
+            conf.write('cutoff 12.\n')
+            conf.write('switching on\n')
+            conf.write('switchdist 10.\n')
+            conf.write('pairlistdist 14\n')
+            conf.write('\n')
+            conf.write('# Integrator Parameters\n')
+            conf.write('timestep 1.0  ;# 1fs/step\n')
+            conf.write('nonbondedFreq 1\n')
+            conf.write('fullElectFrequency 2\n')
+            conf.write('stepspercycle 10\n')
+            conf.write('\n')
+            conf.write('# Output\n')
+            conf.write(f'outputName {job.name}_{job.next_stage}_{steps}\n')
+            conf.write('outputEnergies 100\n')
+            conf.write('outputPressure 100\n')
+            conf.write('\n')
+            conf.write('# Run\n')
+            conf.write(f'minimize {steps}')
+
+    def min2(self, job):
+        #self.solv_ion(job)
+        cbvx, cbvy, cbvz, corx, cory, corz = self.calc_pcell(job)
+        steps = 15000
+        with open(job.conf, 'w') as conf:
+            conf.write('# Input\n')
+            conf.write(f'structure {job.psf}\n')
+            conf.write(f'coordinates {job.pdb}\n')
+            conf.write('paraTypeCharmm on\n')
+            for ff in job.ffs:
+              conf.write(f'parameters {ff}\n')
+            conf.write('\n')
+            conf.write('# Temperature\n')
+            conf.write('temperature 0\n')
+            conf.write('\n')
+            conf.write('# Periodic Boundary Conditions\n')
+            conf.write('wrapWater on\n')
+            conf.write('wrapAll on\n')
+            conf.write(f'cellOrigin\t{corx}\t{cory}\t{corz}\n')
+            conf.write(f'cellBasisVector1\t{cbvx}\t0.0\t0.0\n')
+            conf.write(f'cellBasisVector2\t0.0\t{cbvy}\t0.0\n')
+            conf.write(f'cellBasisVector3\t0.0\t0.0\t{cbvz}\n')
+            conf.write('\n')
+            conf.write('# Force-Field Parameters\n')
+            conf.write('exclude scaled1-4\n')
+            conf.write('1-4scaling 1.0\n')
+            conf.write('cutoff 12.\n')
+            conf.write('switching on\n')
+            conf.write('switchdist 10.\n')
+            conf.write('pairlistdist 14\n')
+            conf.write('\n')
+            conf.write('# Integrator Parameters\n')
+            conf.write('timestep 1.0  ;# 1fs/step\n')
+            conf.write('nonbondedFreq 1\n')
+            conf.write('fullElectFrequency 2\n')
+            conf.write('stepspercycle 10\n')
+            conf.write('\n')
+            conf.write('# Output\n')
+            conf.write(f'outputName {job.name}_{job.next_stage}_{steps}\n')
+            conf.write('outputEnergies 100\n')
+            conf.write('outputPressure 100\n')
+            conf.write('\n')
+            conf.write('# Run\n')
+            conf.write(f'minimize {steps}')
+
+    def heat(self, job):
+        temp_reinit_steps = 100
+        steps = 10000
+        final_steps = 30 * temp_reinit_steps + steps
+        cbvx, cbvy, cbvz, corx, cory, corz = self.calc_pcell(job)
+        with open(job.conf, 'w') as conf:
+            conf.write('# Input\n')
+            conf.write(f'structure {job.psf}\n')
+            conf.write(f'coordinates {job.pdb}\n')
+            conf.write(f'bincoordinates {job.coor}\n')
+            conf.write('paraTypeCharmm on\n')
+            for ff in job.ffs:
+              conf.write(f'parameters {ff}\n')
+            conf.write('\n')
+            conf.write('# Temperature\n')
+            conf.write('temperature 0\n')
+            conf.write('\n')
+            conf.write('# Periodic Boundary Conditions\n')
+            conf.write('wrapWater on\n')
+            conf.write('wrapAll on\n')
+            conf.write(f'cellOrigin\t{corx}\t{cory}\t{corz}\n')
+            conf.write(f'cellBasisVector1\t{cbvx}\t0.0\t0.0\n')
+            conf.write(f'cellBasisVector2\t0.0\t{cbvy}\t0.0\n')
+            conf.write(f'cellBasisVector3\t0.0\t0.0\t{cbvz}\n')
+            conf.write('# Force-Field Parameters\n')
+            conf.write('exclude scaled1-4\n')
+            conf.write('1-4scaling 1.0\n')
+            conf.write('cutoff 12.\n')
+            conf.write('switching on\n')
+            conf.write('switchdist 10.\n')
+            conf.write('pairlistdist 14\n')
+            conf.write('\n')
+            conf.write('# Full Electrostatics\n')
+            conf.write('PME on\n')
+            conf.write('PMEGridSpacing 1.0\n')
+            conf.write('\n')
+            conf.write('# Integrator Parameters\n')
+            conf.write('timestep 1.0 ;# 1fs/step\n')
+            conf.write('nonbondedFreq 1\n')
+            conf.write('fullElectFrequency 2\n')
+            conf.write('stepspercycle 10\n')
+            conf.write('\n')
+            conf.write('# Output\n')
+            conf.write(f'outputName {job.name}_{job.next_stage}_{final_steps}\n')
+            conf.write('outputEnergies 100\n')
+            conf.write('outputPressure 100\n')
+            conf.write('dcdfreq 1000\n')
+            conf.write('\n')
+            conf.write('# Constant Temperature Control\n')
+            conf.write('langevin on ;# do langevin dynamics\n')
+            conf.write('langevinDamping 0.5 ;# damping coefficient (gamma) of 0.5/ps\n')
+            conf.write('langevinTemp 310\n')
+            conf.write('langevinHydrogen yes ;# couple langevin bath to hydrogens\n')
+            conf.write('\n')
+            conf.write('# Constant Pressure Control\n')
+            conf.write('useGroupPressure no ;# needed for 2fs steps\n')
+            conf.write('useFlexibleCell yes ;# no for water box, yes for membrane\n')
+            conf.write('useConstantRatio yes ;# no for water box, yes for membrane\n')
+            conf.write('langevinPiston on\n')
+            conf.write('langevinPistonTarget 1.01325 ;#  in bar -> 1 atm\n')
+            conf.write('langevinPistonPeriod 100.\n')
+            conf.write('langevinPistonDecay 50.\n')
+            conf.write('langevinPistonTemp 310\n')
+            conf.write('\n')
+            conf.write('# Run equilibration\n')
+            conf.write(f'set freq {temp_reinit_steps}\n')
+            conf.write('for {set i 10} {$i <= 310} {incr i 10} {\n')
+            conf.write('reinitvels $i\n')
+            conf.write('langevinTemp $i\n')
+            conf.write('run $freq\n')
+            conf.write('}\n')
+            conf.write('# Run stabilization\n')
+            conf.write(f'run {steps}')
+
+
+    def sim(self, job):
+        cbvx, cbvy, cbvz, corx, cory, corz = self.calc_pcell(job)
+        with open(job.conf, 'w') as conf:
+            conf.write('# Input\n')
+            conf.write(f'structure {job.psf}\n')
+            conf.write(f'coordinates {job.pdb}\n')
+            conf.write(f'bincoordinates {job.coor}\n')
+            conf.write('paraTypeCharmm on\n')
+            for ff in job.ffs:
+              conf.write(f'parameters {ff}\n')
+            conf.write('\n')
+            conf.write('# Temperature\n')
+            conf.write('temperature 0\n')
+            conf.write('\n')
+            conf.write('# Periodic Boundary Conditions\n')
+            conf.write('wrapWater on\n')
+            conf.write('wrapAll on\n')
+            conf.write(f'cellOrigin\t{corx}\t{cory}\t{corz}\n')
+            conf.write(f'cellBasisVector1\t{cbvx}\t0.0\t0.0\n')
+            conf.write(f'cellBasisVector2\t0.0\t{cbvy}\t0.0\n')
+            conf.write(f'cellBasisVector3\t0.0\t0.0\t{cbvz}\n')
+            conf.write('# Force-Field Parameters\n')
+            conf.write('exclude scaled1-4\n')
+            conf.write('1-4scaling 1.0\n')
+            conf.write('cutoff 12.\n')
+            conf.write('switching on\n')
+            conf.write('switchdist 10.\n')
+            conf.write('pairlistdist 14\n')
+            conf.write('\n')
+            conf.write('# Full Electrostatics\n')
+            conf.write('PME on\n')
+            conf.write('PMEGridSpacing 1.0\n')
+            conf.write('\n')
+            conf.write('# Integrator Parameters\n')
+            conf.write('timestep 2.0 ;# 2fs/step\n')
+            conf.write('rigidBonds all ;# needed for 2fs steps\n')
+            conf.write('nonbondedFreq 1\n')
+            conf.write('fullElectFrequency 2\n')
+            conf.write('stepspercycle 10\n')
+            conf.write('\n')
+            conf.write('# Output\n')
+            conf.write(f'outputName {job.name}_{job.next_stage}_{job.next_step}\n')
+            conf.write('outputEnergies 10000\n')
+            conf.write('outputPressure 10000\n')
+            conf.write('dcdfreq 10000\n')
+            conf.write('\n')
+            conf.write('# Constant Temperature Control\n')
+            conf.write('langevin on ;# do langevin dynamics\n')
+            conf.write('langevinDamping 0.5 ;# damping coefficient (gamma) of 0.5/ps\n')
+            conf.write('langevinTemp 310\n')
+            conf.write('langevinHydrogen no ;# couple langevin bath to hydrogens\n')
+            conf.write('\n')
+            conf.write('# Constant Pressure Control\n')
+            conf.write('useGroupPressure yes ;# needed for 2fs steps\n')
+            conf.write('useFlexibleCell yes ;# no for water box, yes for membrane\n')
+            conf.write('useConstantRatio yes ;# no for water box, yes for membrane\n')
+            conf.write('langevinPiston on\n')
+            conf.write('langevinPistonTarget 1.01325 ;#  in bar -> 1 atm\n')
+            conf.write('langevinPistonPeriod 100.\n')
+            conf.write('langevinPistonDecay 50.\n')
+            conf.write('langevinPistonTemp 310\n')
+            conf.write('\n')
+            conf.write('# Run\n')
+            conf.write(f'run {job.steps}')
+
+
+    def del_all_mols(self):
+        for mol in vmd.molecule.listall():
+            vmd.molecule.delete(mol)
+
+
+    def solv_ion(self, job):
+        pdb_f = f'{job.prefix}_{job.stage}.pdb'
+        psf_f = f'{job.prefix}.psf'
+        solv_f = f'{job.prefix}_{job.stage}_solv'
+        solv_ion_f = f'{job.prefix}_{job.stage}_solv_ion'
+        molid = vmd.molecule.load('psf', psf_f, 'namdbin', job.coor)
+        vmd.molecule.write(molid, 'pdb', pdb_f)
+        vmd.evaltcl('package require solvate')
+        vmd.evaltcl(f'solvate {job.psf} {pdb_f} -o {solv_f} '
+                f'-s WT -x 13 -y 13 -z 13 +x 13 +y 13 +z 13 -b 2.4')
+        vmd.evaltcl('package require autoionize')
+        vmd.evaltcl(f'autoionize -psf {solv_f}.psf -pdb {solv_f}.pdb -o {solv_ion_f} '
+                f'-sc 0.15')
+        self.del_all_mols()
+
+
+    def calc_pcell(self, job):
+        if job.next_stage == '2-min' or job.next_stage == '3-min':
+            molid = vmd.molecule.load('psf', job.psf, 'pdb', job.pdb)
+        else:
+            molid = vmd.molecule.load('psf', job.psf, 'namdbin', job.coor)
+        all = vmd.atomsel("all", molid=molid)
+        minmax = all.minmax()
+        center = all.center()
+        cbvx = minmax[1][0] - minmax[0][0]
+        cbvy = minmax[1][1] - minmax[0][1]
+        cbvz = minmax[1][2] - minmax[0][2]
+        corx = center[0]
+        cory = center[1]
+        corz = center[2]
+        self.del_all_mols()
+        return cbvx, cbvy, cbvz, corx, cory, corz