# Comon topology name and address top = 'pNRTapo-strip.pdb' # Number of processors np = 10 # Functions required for parallelization def multi_run_wrapper(args): return f(*args) def f(msm,clL,frame,count): print count structure = msm.draw_samples(clL, 1)[frame] print structure f = md.load(T[structure[0][0]], top=top, frame=structure[0][1]) f.save_pdb(str(count)+'.pdb') print count cluster = pickle.load(open(cl,'rb')) clL = cluster.labels_ msm = io.load(msm) synthTrj = msm.sample_discrete(state = None, n_steps = n_samples) T = [] for trj in sorted(glob.glob(Trjs)): T.append(trj) arg = [(msm,clL,msm_mapping_[synthTrj[count]],count) for count in range(len(synthTrj))] p = Pool(np) S = p.map(multi_run_wrapper, arg)
### Performing kinetic Monte Carlo simulations on Markov state models to estimate the sampling time required to reach the ### target state from an arbitrary inital state using SAXS guided adaptive sampling strategy. ### Required packages: numpy, msmbuilder ### @Chuankai Zhao, [email protected] from msmbuilder.utils import io import numpy as np msm = io.load("MSM25.pkl") Steps = [ 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30 ] Num = 10 Maxrun = 100 Rounds = [round + 1 for round in range(Maxrun)] def loadSAXSdata(): file = open("ProteinG_discrepancy.txt") SAXS = np.loadtxt(file) N_states = np.shape(SAXS)[0] SAXS_dict = {} for i in range(N_states): state = int(SAXS[i][0]) intensity = SAXS[i][1] SAXS_dict[state] = intensity return SAXS_dict
from msmbuilder.utils import io
msm = io.load('MSM.pkl')
counts = msm.countsmat_
import numpy as np
from scipy.optimize import minimize
msm.score
msm.score_ll
sklearn.cross_validation.LeaveOneOut
for j in range(0,parallel_traj[i]):
for k in range(0,seq_traj[i]):
if os.path.isfile(address+'round'+str(rounds[i]+1)+'/traj_files/par'+str(j+1)+'_sim'+str(k+1)+'.mdcrd')== True:
f.write('trajin '+address+'round'+str(rounds[i]+1)+'/traj_files/par'+str(j+1)+'_sim'+str(k+1)+'.mdcrd 1 last 6'+'\n')
else:
pass
f.write('watershell :A8S watershell_pyl10_active.dat W1 lower 3.0 upper 5.0'+ '\n')
f.close()
#-----------------------------------------------------------------------------------
# Calculations of Lys and ser distance from ABA in inactive and active trajectories.
from msmbuilder.utils import io
#in shadowfax
lys=io.load('/home/sshukla4/pyl10/msm/rmsd_dist/dist_data/aba_lys_dist.pkl')
ser=io.load('/home/sshukla4/pyl10/msm/rmsd_dist/dist_data/cl2_ser_aba.pkl')
inactive_lys=lys[0:90]
active_lys=lys[536:576]
inactive_ser=ser[0:90]
active_ser=ser[536:576]
#In lab mac
os.chdir('/Users/Saurabh/Dropbox/My_Papers_and_Reports/Papers/ABA_binding/figures/active_inactive/pyl10')
inactive_lys=io.load('inactive_lys_aba_pyl10.pkl')
active_lys=io.load('active_lys_aba_pyl10.pkl')
inactive_ser=io.load('inactive_ser_aba_pyl10.pkl')
active_ser=io.load('active_ser_aba_pyl10.pkl')
'--stride',
dest='stride',
help='Stride of the raw trajectories being subsampled.',
default=None,
type=int,
required=True)
args = parser.parse_args()
return args
### Main program
if __name__ == '__main__':
options = parse_cmdln()
msm_name = options.msm
cl_name = options.cl
cluster = pickle.load(open(cl_name, 'rb'), encoding='latin1')
msm = io.load(msm_name)
cl_label = cluster.labels_
n_samples = options.NS
stride = options.stride
# Read the topology file and the list of MD raw trajectories
top = "/home/amoffet2/msm_network_project/folding/lindorff-larsen_2011_trajs/protein_g-350K/protein_g.pdb"
trajnames = glob.glob(
"/home/amoffet2/msm_network_project/folding/lindorff-larsen_2011_trajs/protein_g-350K/DESRES-Trajectory_NuG2-*-protein/NuG2-*-protein/*.dcd"
)
# Using msmbuilder.draw_samples function to extract 100 structures from each state
selects = msm.draw_samples(cl_label, n_samples)
pickstates(selects, trajnames, top, stride, options.NP)
