def prepare_confs_iso(self):
prefix = 'ad'
confs_dir = self.ldtdir + self.datdir + '/confs/'
isoms_dir = self.ldtdir + self.datdir + '/confs_iso/'
files = os.listdir(confs_dir)
files = [f for f in files if f.rsplit('.', maxsplit=1)[-1] == 'xyz']
#print(len(files))
ds = dict()
of = open(self.ldtdir + self.datdir + '/info_confstoiso_map.dat', 'w')
for i, f in enumerate(files):
#print(confs_dir + f)
X, S, N, C = hdt.readxyz2(confs_dir + f)
S = np.array(S)
idx = sorted(range(len(S)), key=lambda k: S[k])
S = S[np.array(idx)]
for j, x in enumerate(X):
X[j] = x[idx]
id = "".join(S)
if id in ds:
sid = len(ds[id])
of.write(f + ' ' + convert_eformula(S) + ' ' + str(sid) + ' ' + str(X.shape[0]) + '\n')
ds[id].append((X, S))
else:
of.write(f + ' ' + convert_eformula(S) + ' ' + str(0) + ' ' + str(X.shape[0]) + '\n')
ds.update({id: [(X, S)]})
of.close()
Nt = 0
for i in ds.keys():
X = []
S = []
for j in ds[i]:
X.append(j[0])
S.append(j[1])
X = np.vstack(X)
S = list(S[0])
N = X.shape[0]
Nt += N
if N < 96:
#print(type(S), S)
fn = prefix + '_' + convert_eformula(S) + '-' + str(N).zfill(3) + '.xyz'
#print('Writing: ', fn)
hdt.writexyzfile(isoms_dir + '/' + fn, X, S)
else:
Nsplit = int(math.ceil(N/float(96)))
X = np.array_split(X, Nsplit)
for l,x in enumerate(X):
fn = prefix + '_' + convert_eformula(S) + '_' + str(l).zfill(2) + '-' + str(x.shape[0]).zfill(3) + '.xyz'
hdt.writexyzfile(isoms_dir + '/' + fn, x, S)
def TS_sampling(self, tid, TS_infiles, tsparams, gpuid):
activ = aat.MD_Sampler(TS_infiles, self.netdict['cnstfile'],
self.netdict['saefile'],
self.netdict['nnfprefix'],
self.netdict['num_nets'], gpuid)
T = tsparams['T']
sig = tsparams['sig']
Ns = tsparams['n_samples']
n_steps = tsparams['n_steps']
steps = tsparams['steps']
min_steps = tsparams['min_steps']
nm = tsparams['normalmode']
displacement = tsparams['displacement']
difo = open(
self.ldtdir + self.datdir + '/info_tssampler-' + str(tid) + '.nfo',
'w')
for f in TS_infiles:
X = []
ftme_t = 0.0
fail_count = 0
sumsig = 0.0
for i in range(Ns):
x, S, t, stddev, fail, temp = activ.run_md(
f,
T,
steps,
n_steps,
nmfile=f.rsplit(".", 1)[0] + '.log',
displacement=displacement,
min_steps=min_steps,
sig=sig,
nm=nm)
sumsig += stddev
if fail:
#print('Job '+str(i)+' failed in '+"{:.2f}".format(t)+' Sigma: ' + "{:.2f}".format(stddev)+' SetTemp: '+"{:.2f}".format(temp))
difo.write('Job ' + str(i) + ' failed in ' +
"{:.2f}".format(t) + 'fs Sigma: ' +
"{:.2f}".format(stddev) + ' SetTemp: ' +
"{:.2f}".format(temp) + '\n')
X.append(x[np.newaxis, :, :])
fail_count += 1
else:
#print('Job '+str(i)+' succeeded.')
difo.write('Job ' + str(i) + ' succeeded.\n')
ftme_t += t
print('Complete mean fail time: ' +
"{:.2f}".format(ftme_t / float(Ns)) + ' failed ' +
str(fail_count) + '/' + str(Ns) + '\n')
difo.write('Complete mean fail time: ' +
"{:.2f}".format(ftme_t / float(Ns)) + ' failed ' +
str(fail_count) + '/' + str(Ns) + ' MeanSig: ' +
"{:.2f}".format(sumsig / float(Ns)) + '\n')
X = np.vstack(X)
hdt.writexyzfile(self.cdir + os.path.basename(f), X, S)
del activ
difo.close()
def mol_dyn_sampling(self, md_work, i, N, T1, T2, dt, Nc, Ns, sig, gpuid):
activ = aat.moldynactivelearning(self.netdict['cnstfile'],
self.netdict['saefile'],
self.netdict['nnfprefix'],
self.netdict['num_nets'], gpuid)
difo = open(
self.ldtdir + self.datdir + '/info_data_mdso-' + str(i) + '.nfo',
'w')
Nmol = 0
dnfo = 'MD Sampler running: ' + str(md_work.size)
difo.write(dnfo + '\n')
Nmol = md_work.size
ftme_t = 0.0
for di, id in enumerate(md_work):
data = hdt.read_rcdb_coordsandnm(id)
#print(di, ') Working on', id, '...')
S = data["species"]
# Set mols
activ.setmol(data["coordinates"], S)
# Generate conformations
X = activ.generate_conformations(N, T1, T2, dt, Nc, Ns, dS=sig)
ftme_t += activ.failtime
nfo = activ._infostr_
m = id.rsplit('/', 1)[1].rsplit('.', 1)[0]
difo.write(' -' + m + ': ' + nfo + '\n')
difo.flush()
#print(nfo)
if X.size > 0:
hdt.writexyzfile(
self.cdir + 'mds_' + m.split('.')[0] + '_' +
str(i).zfill(2) + str(di).zfill(4) + '.xyz', X, S)
difo.write('Complete mean fail time: ' +
"{:.2f}".format(ftme_t / float(Nmol)) + '\n')
print(Nmol)
del activ
difo.close()
storedir = '/home/jujuman/Research/extensibility_test_sets/drugbank/'
suppl = Chem.SDMolSupplier(
'/home/jujuman/Dropbox/ChemSciencePaper.AER/Benchmark_Datasets/drugbank/drugbank_3d_1564.sdf',
removeHs=False)
for id, m in enumerate(suppl):
if m is None: continue
name = m.GetProp('_Name')
xyz, spc = pya.__convert_rdkitconfs_to_nparr__(m)
print(xyz.shape)
print(name, id, spc)
hdt.writexyzfile(storedir + 'xyz/drugbank_' + str(id).zfill(4) + '.xyz',
xyz, spc)
hdt.write_rcdb_input(xyz[0],
spc,
id,
storedir,
'drugbank',
10,
'wb97x/6-31g*',
'300.0',
fill=4,
comment='Name: ' + name)
# Print size
#print(m.GetNumAtoms(), m.GetNumHeavyAtoms())
def plot_irc(axes, i, d, f):
#print(f)
Eact, xyz, spc, Rc = pyg.read_irc(d + f)
Eact = hdt.hatokcal * Eact
xyz = xyz[1:]
Eact = Eact[1:]
Rc = Rc[:-1]
#print(Rc[:,1])
#print(Eact-Eact.min() - Rc[:,1]-Rc[:,1].min())
s_idx = f.split('IRC')[1].split('.')[0]
hdt.writexyzfile(c + f.split('.')[0] + '.xyz', xyz, spc)
#print(f.split('IRC')[1].split('.')[0],Rc.shape)
if Rc.size > 10:
#------------ CV NETWORKS 1 -----------
energies1 = []
N = 0
for comp in nc1:
comp.setConformers(confs=xyz, types=list(spc))
energies1.append(hdt.hatokcal * comp.energy())
N = N + 1
energies2 = []
N = 0
for comp in nc2:
comp.setConformers(confs=xyz, types=list(spc))
energies2.append(hdt.hatokcal * comp.energy())
N = N + 1
modl_std1 = np.std(energies1, axis=0)[::-1]
energies1 = np.mean(np.vstack(energies1), axis=0)
modl_std2 = np.std(energies2, axis=0)[::-1]
energies2 = np.mean(np.vstack(energies2), axis=0)
rmse1 = hdt.calculaterootmeansqrerror(energies1, Eact)
rmse2 = hdt.calculaterootmeansqrerror(energies2, Eact)
dba = Eact.max() - Eact[0]
db1 = energies1.max() - energies1[0]
db2 = energies2.max() - energies2[0]
rpa = Eact[0] - Eact[-1]
rp1 = energies1[0] - energies1[-1]
rp2 = energies2[0] - energies2[-1]
bar1.append(abs(db1 - dba))
bar2.append(abs(db2 - dba))
rmp1.append(abs(rpa - rp1))
rmp2.append(abs(rpa - rp2))
Ec1.append(energies1)
Ec2.append(energies2)
Ea.append(Eact)
print(i, ')', f, ':', len(spc), ':', rmse1, rmse2, 'R/P1: ',
abs(rpa - rp1), 'R/P2: ', abs(rpa - rp2), 'Barrier1:',
abs(db1 - dba), 'Barrier2:', abs(db2 - dba))
Rce = hdt.hatokcal * (Rc[:, 0] - Rc[:, 0][0])
Rce1 = energies2[::-1] - energies2[::-1][0]
axes.set_xlim([Rc.min(), Rc.max()])
axes.set_ylim([Rce.min() - 1.0, Rce1.max() + 20.0])
axes.plot(Rc[:, 1],
hdt.hatokcal * (Rc[:, 0] - Rc[:, 0][0]),
color='Black',
label='DFT')
axes.errorbar(Rc[:, 1],
energies2[::-1] - energies2[::-1][0],
yerr=modl_std2,
fmt='--',
color='red',
label="ANI-1: " + "{:.1f}".format(bar2[-1]),
linewidth=2)
axes.errorbar(Rc[:, 1],
energies1[::-1] - energies1[::-1][0],
yerr=modl_std1,
fmt='--',
color='blue',
label="[" + str(i) + "]: " + "{:.1f}".format(bar1[-1]),
linewidth=2)
#axes.set_xlabel("Reaction Coordinate $\AA$")
#axes.set_ylabel(r"$\Delta E$ $ (kcal \times mol^{-1})$")
#axes.plot(Rc[:, 1], energies2[::-1]-energies2[::-1][0],'--',color='red',label="["+str(i)+"]: "+"{:.1f}".format(bar2[-1]),linewidth=3)
#axes.plot(Rc[:, 1], energies1[::-1]-energies1[::-1][0],'--',color='green',label="["+str(i)+"]: "+"{:.1f}".format(bar1[-1]),linewidth=3)
axes.legend(loc="upper left", fontsize=10)
axes.set_title(str(f),
color='black',
fontdict={'weight': 'bold'},
x=0.8,
y=0.85)
El = np.vstack(El)
#dEm = np.sum(El, axis=0) / 3.0
bad_idx = []
god_idx = []
for i in range(Ea.shape[0]):
test = np.array([True if j > 1.0 else False for j in El[:,i]])
if np.all(test):
bad_idx.append(i)
else:
god_idx.append(i)
Nd = Nd + len(god_idx)
Nb = Nb + len(bad_idx)
if len(god_idx) != 0:
# print(gn)
dpack.store_data(data['parent'] + "/" + data['name'], coordinates=X[god_idx], energies=Ea[god_idx], species=S)
if len(bad_idx) > 0:
print(data['parent'],data['name'],'- Bad:',len(bad_idx),'of',len(god_idx))
for i in bad_idx:
print('(', i, ')', ':' ,El[:,i])
hdn.writexyzfile(store_xyz+data['parent'] + '_' + data['name']+'.xyz',X[bad_idx],S)
dpack.cleanup()
print('Bad:',Nb,'of',Nd+Nb)
EANI = np.asarray(EANI).flatten()
# Max
if abs(E.max()) > maxi:
maxi = abs(E.max())
# Min
if abs(E.min()) < mini:
mini = abs(E.min())
#
df_E.append(E)
df_Ea.append(EANI)
#print (df_E)
hdn.writexyzfile("coordlist.xyz", np.concatenate(coord_list), spec_list[0])
import matplotlib.pyplot as plt
df_E = hdn.hatokcal * np.array(df_E).flatten()
df_Ea = hdn.hatokcal * np.array(df_Ea).flatten()
rmse = hdn.calculaterootmeansqrerror(df_E, df_Ea)
x = list(range(0, df_E.shape[0]))
plt.scatter(df_D, df_E, label='DFT')
plt.scatter(df_D,
df_Ea,
color='red',
label='ANI RMSE: ' + "{:.2f}".format(rmse) + 'kcal/mol',
sid = len(ds[id])
of.write(f + ' ' + convert_eformula(S) + ' ' + str(sid) + ' ' +
str(X.shape[0]) + '\n')
ds[id].append((X, S))
else:
of.write(f + ' ' + convert_eformula(S) + ' ' + str(0) + ' ' +
str(X.shape[0]) + '\n')
ds.update({id: [(X, S)]})
of.close()
#print(i,len(ds))
Nt = 0
for i in ds.keys():
X = []
S = []
for j in ds[i]:
X.append(j[0])
S.append(j[1])
X = np.vstack(X)
S = list(S[0])
N = X.shape[0]
Nt += N
print(type(S), S)
fn = prefix + '_' + convert_eformula(S) + '-' + str(N).zfill(5) + '.xyz'
print('Writing: ', fn)
hdt.writexyzfile(ndir + fn, X, S)
print('Total data:', Nt)
def __fragmentbox__(self, file, sig_cut):
self.X = self.mol.get_positions()
self.frag_list = []
self.Nd = 0
self.Nt = 0
for i in range(len(self.Na)):
si = self.ctd[i][2]
di = self.ctd[i][1]
Nai = self.Na[i]
Xci = np.sum(self.X[si:si + Nai, :], axis=0) / Nai
Xi = self.X[si:si + Nai, :]
if np.all(Xci > di) and np.all(Xci < self.L - di):
for j in range(i + 1, len(self.Na)):
sj = self.ctd[j][2]
dj = self.ctd[j][1]
Naj = self.Na[j]
Xcj = np.sum(self.X[sj:sj + Naj, :], axis=0) / Naj
Xj = self.X[sj:sj + Naj, :]
if np.all(Xcj > dj) and np.all(Xcj < self.L - dj):
dc = np.linalg.norm(Xci - Xcj)
if dc < di + dj + 6.0:
min = 10.0
for ii in range(Nai):
Xiii = Xi[ii]
for jj in range(Naj):
Xjjj = Xj[jj]
v = np.linalg.norm(Xiii - Xjjj)
if v < min:
min = v
if min < 4.2 and min > 1.1:
Xf = np.vstack([Xi, Xj])
Sf = self.S[si:si + Nai]
Sf.extend(self.S[sj:sj + Naj])
Xcf = np.sum(Xf, axis=0) / (Nai + Naj)
Xf = Xf - Xcf
E = np.empty(5, dtype=np.float64)
for id, nc in enumerate(self.aens.ncl):
nc.setMolecule(coords=np.array(
Xf, dtype=np.float32),
types=Sf)
E[id] = nc.energy()[0]
sig = np.std(
hdn.hatokcal * E) / np.sqrt(Nai + Naj)
self.Nt += 1
if sig > sig_cut:
self.Nd += 1
hdn.writexyzfile(
file + str(i).zfill(4) + '-' +
str(j).zfill(4) + '.xyz',
Xf.reshape(1, Xf.shape[0], 3), Sf)
self.frag_list.append(
dict({
'coords': Xf,
'spec': Sf
}))
mol.set_calculator(ANI(False))
mol.calc.setnc(nc)
# Optimize molecule
start_time = time.time()
dyn = LBFGS(mol)
dyn.run(fmax=C)
print('[ANI Total time:', time.time() - start_time, 'seconds]')
print(hdt.evtokcal * mol.get_potential_energy())
# Save optimized mol
spc = mol.get_chemical_symbols()
pos = mol.get_positions(wrap=True).reshape(1, len(spc), 3)
hdt.writexyzfile(optfile, pos, spc)
exit(0)
# Open MD output
mdcrd = open(xyzfile, 'w')
# Open MD output
traj = open(trajfile, 'w')
# We want to run MD with constant energy using the Langevin algorithm
# with a time step of 0.5 fs, the temperature T and the friction
# coefficient to 0.02 atomic units.
dyn = Langevin(mol, 0.1 * units.fs, T * units.kB, 0.005)
# Run equilibration
import hdnntools as hdt
import pyanitools as pyt
import os
file = '/home/jujuman/Research/DataReductionMethods/model6/model0.05me/ani_red_c06.h5'
sdir = '/home/jujuman/Research/GDB-11-AL-wB97x631gd/'
aload = pyt.anidataloader(file)
for data in aload:
X = data['coordinates']
S = data['species']
P = data['path']
parent = P.split('/')[1]
index = P.split('/')[2].split('mol')[1].zfill(7)
path = sdir+parent
if not os.path.exists(path):
os.mkdir(path)
print(path + '/' + parent + '-' + index + '.xyz','DATA:',X.shape[0])
hdt.writexyzfile(path+'/'+parent+'-'+index+'.xyz',X,S)
xyz = data["coordinates"]
nmc = data["nmdisplacements"]
frc = data["forceconstant"]
nms = nmt.nmsgenerator(xyz, nmc, frc, spc, T, minfc=1.0E-2)
conformers = []
for i in range(Ngen):
conformers.append(nms.get_random_structure())
conformers = np.stack(conformers)
nc.setConformers(confs=conformers, types=list(spc))
Ecmp = nc.energy() # this generates AEVs
aevs = np.empty([Ngen, len(rcatoms) * aevsize])
for m in range(Ngen):
for j, a in enumerate(rcatoms):
aevs[m, j * aevsize:(j + 1) * aevsize] = nc.atomicenvironments(
a, m).copy()
dm = scispc.distance.pdist(aevs, 'sqeuclidean')
picker = rdSimDivPickers.MaxMinPicker()
seed_list = [i for i in range(Ngen)]
np.random.shuffle(seed_list)
print('seed:', seed_list)
ids = list(picker.Pick(dm, Ngen, Nkep, firstPicks=list(seed_list[0:5])))
ids.sort()
print(f, len(ids), conformers.shape, dm.shape, ":", ids)
hdn.writexyzfile(cdir + f.split('.')[0] + '.xyz', conformers[ids], spc)
of.close()
psi.append(i.split('_')[2].split('.')[0])
data = hdt.readxyz2(dir + i)
print(data[0])
print(data[1])
xyz.append(data[0])
nc.setMolecule(coords=data[0][0], types=list(data[1]))
Eact.append(nc.energy()[0])
xyz = np.vstack(xyz)
#print(xyz)
hdt.writexyzfile(
'/home/jujuman/Dropbox/ChemSciencePaper.AER/JustinsDocuments/ACS_april_2017/DipeptidePhiPsi/data.xyz',
xyz, list(data[1]))
data2 = hdt.readncdat(
'/home/jujuman/Dropbox/ChemSciencePaper.AER/JustinsDocuments/ACS_april_2017/DipeptidePhiPsi/data.dat',
type=np.float32)
Eact = np.array(Eact)
Edft = data2[2]
z = np.array(hdt.hatokcal * (Eact - Eact.min()),
dtype=np.float32).reshape(x.shape[0], x.shape[0])
z2 = np.array(hdt.hatokcal * (Edft - Edft.min()),
dtype=np.float32).reshape(x.shape[0], x.shape[0])
rmse = hdt.calculaterootmeansqrerror(z, z2)
def pDyn_QMsampling(self, pdynparams, gpuid):
#Call subproc_pDyn class in pyaniasetools as activ
activ = aat.subproc_pDyn(self.netdict['cnstfile'],
self.netdict['saefile'],
self.netdict['nnfprefix'],
self.netdict['num_nets'], gpuid)
pDyn_dir = pdynparams['pDyn_dir'] #Folder to write pDynamo input file
num_rxn = pdynparams['num_rxn'] #Number of input rxn
logfile_OPT = pdynparams['logfile_OPT'] #logfile for FIRE OPT output
logfile_TS = pdynparams['logfile_TS'] #logfile for ANI TS output
logfile_IRC = pdynparams['logfile_IRC'] #logfile for ANI IRC output
sbproc_cmdOPT = pdynparams[
'sbproc_cmdOPT'] #Subprocess commands to run pDyanmo
sbproc_cmdTS = pdynparams['sbproc_cmdTS']
sbproc_cmdIRC = pdynparams['sbproc_cmdIRC']
IRCdir = pdynparams['IRCdir'] #path to get pDynamo saved IRC points
indir = pdynparams[
'indir'] #path to save XYZ files of IRC points to check stddev
XYZfile = pdynparams[
'XYZfile'] #XYZ file with high standard deviations structures
l_val = pdynparams[
'l_val'] #Ri --> randomly perturb in the interval [+x,-x]
h_val = pdynparams['h_val']
n_points = pdynparams[
'n_points'] #Number of points along IRC (forward+backward+1 for TS)
sig = pdynparams['sig']
N = pdynparams['N']
wkdir = pdynparams['wkdir']
cnstfilecv = pdynparams['cnstfilecv']
saefilecv = pdynparams['saefilecv']
Nnt = pdynparams['Nnt']
# --------------------------------- Run pDynamo ---------------------------
# auto-TS ---> FIRE constraint OPT of core C atoms ---> ANI TS ---> ANI IRC
activ.write_pDynOPT(num_rxn, pDyn_dir, wkdir, cnstfilecv, saefilecv,
Nnt) #Write pDynamo input file in pDyndir
activ.write_pDynTS(num_rxn, pDyn_dir, wkdir, cnstfilecv, saefilecv,
Nnt)
activ.write_pDynIRC(num_rxn, pDyn_dir, wkdir, cnstfilecv, saefilecv,
Nnt)
chk_OPT = activ.subprocess_cmd(sbproc_cmdOPT, False, logfile_OPT)
if chk_OPT == 0: #Wait until previous subproc is done!!
chk_TS = activ.subprocess_cmd(sbproc_cmdTS, False, logfile_TS)
if chk_TS == 0:
chk_IRC = activ.subprocess_cmd(sbproc_cmdIRC, False,
logfile_IRC)
# ----------------------- Save points along ANI IRC ------------------------
IRCfils = os.listdir(IRCdir)
IRCfils.sort()
for f in IRCfils:
activ.getIRCpoints_toXYZ(n_points, IRCdir + f, f, indir)
infils = os.listdir(indir)
infils.sort()
# ------ Check for high standard deviation structures and get vib modes -----
for f in infils:
stdev = activ.check_stddev(indir + f, sig)
if stdev > sig: #if stddev is high then get modes for that point
nmc = activ.get_nm(indir +
f) #save modes in memory for later use
activ.write_nm_xyz(
XYZfile
) #writes all the structures with high standard deviations to xyz file
# ----------------------------- Read XYZ for NM -----------------------------
X, spc, Na, C = hdt.readxyz3(XYZfile)
# --------- NMS for XYZs with high stddev --------
for i in range(len(X)):
for j in range(len(nmc)):
gen = nmt.nmsgenerator_RXN(
X[i], nmc[j], spc[i], l_val,
h_val) # xyz,nmo,fcc,spc,T,Ri_-x,Ri_+x,minfc = 1.0E-3
N = 500
gen_crd = np.zeros((N, len(spc[i]), 3), dtype=np.float32)
for k in range(N):
gen_crd[k] = gen.get_random_structure()
hdt.writexyzfile(self.cdir + 'nms_TS%i.xyz' % N, gen_crd,
spc[i])
del activ
dtype=np.float32).reshape(xyz.shape[0], 3)
#if conv:
# print('Failed to converge!!!')
energies = np.zeros((Nnc), dtype=np.float64)
#------------ CV NETWORKS 1 -----------
N = 0
for comp in nc1:
comp.setMolecule(coords=xyz, types=list(spc))
energies[N] = comp.energy()[0]
N = N + 1
if np.std(hdt.hatokcal * energies) > 5.0:
hdt.writexyzfile(
'/home/jujuman/Research/CrossValidation/GDB-06-High-sdev/CV1bmol-'
+ str(total_mol) + '.xyz',
xyz.reshape(1, xyz.shape[0], xyz.shape[1]), spc)
total_bad = total_bad + 1
perc = int(100.0 * total_bad / float(total_mol))
output = ' ' + str(k) + ' ' + str(total_bad) + '/' + str(
total_mol) + ' ' + str(perc) + '% (' + str(
Na) + ') : stps=' + str(stps) + ' : ' + str(
energies) + ' : std(kcal/mol)=' + str(
np.std(hdt.hatokcal *
energies)) + ' : ' + Chem.MolToSmiles(m)
if np.std(hdt.hatokcal * energies) > 5.0:
print("CV1:", output)
f1.write(output + '\n')
if not ('F' in Chem.MolToSmiles(m)):
print(n,') Working on',l,Chem.MolToSmiles(m),'...')
# Add hydrogens
m = Chem.AddHs(m)
# generate Nc conformers
cids = AllChem.EmbedMultipleConfs(m, N, useRandomCoords=True)
# Classical Optimization
for cid in cids:
_ = AllChem.MMFFOptimizeMolecule(m, confId=cid, maxIters=250)
# Set mols
activ.setrdkitmol(m,cids)
# Generate conformations
X = activ.generate_conformations(N, T, dt, 2000, 10, dS = 0.34)
nfo = activ._infostr_
difo.write(' -'+l+': '+nfo+'\n')
print(nfo)
difo.flush()
# Set chemical symbols
Xi, S = pya.__convert_rdkitmol_to_nparr__(m)
# Store structures
hdt.writexyzfile(sdir+l+'-'+str(n).zfill(2)+'.xyz', X, S)
difo.close()
Nnc = 2
files = os.listdir(d)
files.sort()
#print(files)
# Construct pyNeuroChem classes
nc1 = [pync.conformers(wkdir1 + cnstfile, wkdir1 + saefile, wkdir1 + 'train' + str(l) + '/networks/', 0, False) for l in range(Nnc)]
comp_xyz = []
for f in files:
#print(f)
Eact, xyz, spc, Rc = pyg.read_irc(d+f)
s_idx = f.split('IRC')[1].split('.')[0]
hdt.writexyzfile(c+f.split('.')[0]+'.xyz',xyz,spc)
#print(f.split('IRC')[1].split('.')[0],Rc.shape)
if Rc.size > 10:
#------------ CV NETWORKS 1 -----------
energies = []
N = 0
for comp in nc1:
comp.setConformers(confs=xyz, types=list(spc))
energies.append(hdt.hatokcal*comp.energy())
N = N + 1
energies = np.vstack(energies)
modl_std = np.std(energies[::-1],axis=0) / float(len(spc))
bad_cnt = 0
bad_xyz = []
nms = nmt.nmsgenerator(xyz,nmc,frc,spc,T,minfc=5.0E-2)
conformers = nms.get_Nrandom_structures(Ngen)
ids = dc.get_divconfs_ids(conformers, spc, Ngen, Nkep, atmlist)
conformers = conformers[ids]
print(' -',f,len(ids),conformers.shape)
sigma = anicv.compute_stddev_conformations(conformers,spc)
#print(sigma)
sid = np.where( sigma > 0.0 )[0]
print(' -', fi, 'of', len(files), ') File:', f, 'keep:', sid.size,'percent:',"{:.2f}".format(100.0*sid.size/Ngen))
Nt += sigma.size
Nk += sid.size
if 100.0*sid.size/float(Ngen) > 0:
Nkp += sid.size
cfn = f.split('.')[0].split('-')[0]+'_'+str(idx).zfill(5)+'-'+f.split('.')[0].split('-')[1]+'.xyz'
hdn.writexyzfile(cdir+cfn,conformers[sid],spc)
idx += 1
Nkt += Nk
Ntt += Nt
of.write(' -Total: '+str(Nk)+' of '+str(Nt)+' percent: '+"{:.2f}".format(100.0*Nk/Nt)+'\n')
of.flush()
print(' -Total:',Nk,'of',Nt,'percent:',"{:.2f}".format(100.0*Nk/Nt))
of.write('\nGrand Total: '+ str(Nkt)+ ' of '+ str(Ntt)+' percent: '+"{:.2f}".format(100.0*Nkt/Ntt)+ ' Kept '+str(Nkp)+'\n')
print('\nGrand Total:', Nkt, 'of', Ntt,'percent:',"{:.2f}".format(100.0*Nkt/Ntt), 'Kept',Nkp)
of.close()
dtdir = '/home/jujuman/Dropbox/IRC_DBondMig/Benzene_rxn2/rxn_ben'+str(n)+'/IRC.log'
#en, cd, ty = pyg.get_irc_data(dtdir+'IRC_fwd.log',dtdir+'IRC_bck.log',dtdir+'saddle_ts.log')
en, cd, ty, Rc = pyg.read_irc(dtdir)
np.savez(wkdir+'reaction_coordinate.npz',x=Rc)
Na = len(ty)
Nc = en.shape[0]
print (cd.shape,' ',en.shape)
#for i,x in enumerate(cd):
#hdt.write_rcdb_input(x,ty,i,wkdir,fpf,5,'wb97x/6-31g*','600.0',opt='0')
hdt.writexyzfile(wkdir+'irc.xyz',cd,ty)
f = open(wkdir+'irc.dat','w')
f.write("comment\n")
f.write(str(Nc)+'\n')
f.write(str(Na)+',')
for j in ty: f.write(j+',')
f.write('\n')
mol = 0
for l,i in enumerate(cd):
for j in i:
for k in j:
f.write(str(k)+',')
f.write('')
f.write(str(en[l]) + ',' + '\n')
import hdnntools as hdn
import pyNeuroChem as pync
import matplotlib.pyplot as plt
file = "/home/jujuman/Research/ANI-DATASET/rxn_db_mig.h5"
al = pyt.anidataloader(file)
al.totalload()
data = al.get_all_data()
al.cleanup()
df_E = hdn.hatokcal * data[1].flatten()
xyz = data[0].reshape(df_E.shape[0], len(data[2][0]), 3)
hdn.writexyzfile('/home/jujuman/crds.xyz', xyz, data[2][0])
# Set required files for pyNeuroChem
#wkdir = '/home/jujuman/Dropbox/ChemSciencePaper.AER/ANI-c08e-ntwk/'
wkdir = '/home/jujuman/Research/GDB-11-wB97X-6-31gd/ANI-c08e-ntwk_newtrain/'
cnstfile = wkdir + 'rHCNO-4.6A_16-3.1A_a4-8.params'
saefile = wkdir + 'sae_6-31gd.dat'
nnfdir = wkdir + 'networks/'
# Construct pyNeuroChem class
mol = pync.conformers(cnstfile, saefile, nnfdir, 0)
mol.setConformers(confs=xyz, types=list(data[2][0]))
E = hdn.hatokcal * mol.energy()
#Nd += len(X)
#Nt += Nu
Nt += 1
if len(X) > 0:
Nd += 1
X = np.stack(X)
x = X[0].reshape(1, X.shape[1], 3) # keep only the first guy
print(' -kept', len(x), 'of', Nu)
if len(X) > 0:
S = gdb.get_symbols_rdkitmol(m)
hdn.write_rcdb_input(x[0],
S,
int(id),
wdir,
fpf,
100,
LOT,
'500.0',
fill=4,
comment='smiles: ' + Chem.MolToSmiles(m) +
' GDB-ID: ' + str(ridx))
hdn.writexyzfile(wdir + fpf + '-' + str(id).zfill(4) + '.xyz',
x.reshape(1, x.shape[1], x.shape[2]), S)
#print(str(id).zfill(8))
molnfo.close()
print('Total mols:', Nd, 'of', Nt, 'percent:',
"{:.2f}".format(100.0 * Nd / float(Nt)))
mol = Atoms(spc, xyz, calculator=ANI(False))
mol.calc.setnc(nc)
LBFGS(mol).run(fmax=0.00001)
vib = Vibrations(mol)
vib.run()
vib.summary()
print(xyz)
xyz = mol.get_positions().copy()
print(xyz)
nm_cr = vib.modes[6:]
Nf = 3 * len(spc) - 6
nmo = nm_cr.reshape(Nf, len(spc), 3)
fcc = np.array([
1.314580, 1.3147106, 1.3149728, 1.5161799, 1.5164505, 5.6583018, 6.7181139,
6.7187967, 6.7193842
])
gen = nm.nmsgenerator(xyz, nmo, fcc, spc, 2000.0)
N = 2000
gen_crd = np.zeros((N, len(spc), 3), dtype=np.float32)
for i in range(N):
gen_crd[i] = gen.get_random_structure()
hdt.writexyzfile('pynmstesting.xyz', gen_crd, spc)
#dir = '/home/jujuman/Research/GDB-11-wB97X-6-31gd/dnnts_rxns/scans_double_bond_migration/' #dir = '/home/jujuman/Dropbox/IRC_DBondMig/rxn1/' dir = '/home/jujuman/Dropbox/IRC_DBondMig/rxn5/' filef = dir + 'IRC_fwd.log' fileb = dir + 'IRC_bck.log' dataf = g09.read_irc(filef) datab = g09.read_irc(fileb) xyz = np.concatenate([np.flipud(datab[1]), dataf[1]]) eng = hdn.hatokcal * np.concatenate([np.flipud(datab[0]), dataf[0]]) print(xyz.shape) hdn.writexyzfile(dir + 'scan.xyz', xyz, dataf[2]) # Set required files for pyNeuroChem # Set required files for pyNeuroChem anipath = '/home/jujuman/Dropbox/ChemSciencePaper.AER/ANI-c08e-ntwk' cnstfile = anipath + '/rHCNO-4.6A_16-3.1A_a4-8.params' saefile = anipath + '/sae_6-31gd.dat' nnfdir = anipath + '/networks/' # Construct pyNeuroChem class nc = pync.conformers(cnstfile, saefile, nnfdir, 0) # Set the conformers in NeuroChem nc.setConformers(confs=xyz, types=dataf[2])
ftme_t = 0.0
for n, m in enumerate(files[0:int(id[0] * len(files))]):
data = hdn.read_rcdb_coordsandnm(id[1] + m)
S = data["species"]
print(n, ') Working on', m, '...')
# Set mols
activ.setmol(data["coordinates"], S)
# Generate conformations
X = activ.generate_conformations(N, T1, T2, dt, Nc, Ns, dS=0.25)
ftme += activ.failtime
ftme_t += activ.failtime
nfo = activ._infostr_
difo.write(' -' + m + ': ' + nfo + '\n')
difo.flush()
print(nfo)
if X.size > 0:
hdn.writexyzfile(
dstore + 'mds_' + m.split('.')[0] + '_' + str(di).zfill(4) +
'.xyz', X, S)
difo.write('Class mean fail time: ' +
"{:.2f}".format(ftme_t / float(len(files))) + '\n')
difo.write('Complete mean fail time: ' + "{:.2f}".format(ftme / float(Nmol)) +
'\n')
print(Nmol)
difo.close()
def normal_mode_sampling(self, T, Ngen, Nkep, sig, gpuid):
of = open(self.ldtdir + self.datdir + '/info_data_nms.nfo', 'w')
aevsize = self.netdict['aevsize']
anicv = aat.anicrossvalidationconformer(self.netdict['cnstfile'],
self.netdict['saefile'],
self.netdict['nnfprefix'],
self.netdict['num_nets'],
gpuid, False)
dc = aat.diverseconformers(self.netdict['cnstfile'],
self.netdict['saefile'],
self.netdict['nnfprefix'] + '0/networks/',
aevsize, gpuid, False)
Nkp = 0
Nkt = 0
Ntt = 0
idx = 0
for di, id in enumerate(self.idir):
of.write(
str(di) + ' of ' + str(len(self.idir)) + ') dir: ' + str(id) +
'\n')
#print(di,'of',len(self.idir),') dir:', id)
files = os.listdir(id)
files.sort()
Nk = 0
Nt = 0
for fi, f in enumerate(files):
data = hdt.read_rcdb_coordsandnm(id + f)
#print(id+f)
spc = data["species"]
xyz = data["coordinates"]
nmc = data["nmdisplacements"]
frc = data["forceconstant"]
nms = nmt.nmsgenerator(xyz, nmc, frc, spc, T, minfc=5.0E-2)
conformers = nms.get_Nrandom_structures(Ngen)
ids = dc.get_divconfs_ids(conformers, spc, Ngen, Nkep, [])
conformers = conformers[ids]
#print(' -',f,len(ids),conformers.shape)
sigma = anicv.compute_stddev_conformations(conformers, spc)
#print(sigma)
sid = np.where(sigma > sig)[0]
#print(sid)
#print(' -', fi, 'of', len(files), ') File:', f, 'keep:', sid.size,'percent:',"{:.2f}".format(100.0*sid.size/Ngen))
Nt += sigma.size
Nk += sid.size
if 100.0 * sid.size / float(Ngen) > 0:
Nkp += sid.size
cfn = f.split('.')[0].split('-')[0] + '_' + str(idx).zfill(
5) + '-' + f.split('.')[0].split('-')[1] + '_2.xyz'
hdt.writexyzfile(self.cdir + cfn, conformers[sid], spc)
idx += 1
Nkt += Nk
Ntt += Nt
of.write(' -Total: ' + str(Nk) + ' of ' + str(Nt) +
' percent: ' + "{:.2f}".format(100.0 * Nk / Nt) + '\n')
of.flush()
#print(' -Total:',Nk,'of',Nt,'percent:',"{:.2f}".format(100.0*Nk/Nt))
del anicv
del dc
of.write('\nGrand Total: ' + str(Nkt) + ' of ' + str(Ntt) +
' percent: ' + "{:.2f}".format(100.0 * Nkt / Ntt) + ' Kept ' +
str(Nkp) + '\n')
#print('\nGrand Total:', Nkt, 'of', Ntt,'percent:',"{:.2f}".format(100.0*Nkt/Ntt), 'Kept',Nkp)
of.close()
def __fragmentbox__(self, file, sighat):
self.X = self.mol.get_positions()
self.frag_list = []
self.Nd = 0
self.Nt = 0
self.maxsig = 0
for i in range(len(self.Na)):
si = self.ctd[i][2]
di = self.ctd[i][1]
Nai = self.Na[i]
Xci = np.sum(self.X[si:si + Nai, :], axis=0) / Nai
Xi = self.X[si:si + Nai, :]
if np.all(Xci > 4.5) and np.all(Xci <= self.L - 4.5):
if np.all(Xci > di) and np.all(Xci < self.L - di):
Xf = Xi
Sf = self.S[si:si + Nai]
Nmax = random.randint(2, 14)
Nmol = 0
for j in range(len(self.Na)):
if i != j:
sj = self.ctd[j][2]
dj = self.ctd[j][1]
Naj = self.Na[j]
Xcj = np.sum(self.X[sj:sj + Naj, :], axis=0) / Naj
Xj = self.X[sj:sj + Naj, :]
if np.all(Xcj > dj) and np.all(Xcj < self.L - dj):
dc = np.linalg.norm(Xci - Xcj)
if dc < di + dj + 5.0:
min = 10.0
for ii in range(Nai):
Xiii = Xi[ii]
for jj in range(Naj):
Xjjj = Xj[jj]
v = np.linalg.norm(Xiii - Xjjj)
if v < min:
min = v
if min < 4.5 and min > 0.70:
Xf = np.vstack([Xf, Xj])
Sf.extend(self.S[sj:sj + Naj])
Nmol += 1
if Nmol >= Nmax:
break
Xcf = np.sum(Xf, axis=0) / float(len(Sf))
Xf = Xf - Xcf
E = np.empty(5, dtype=np.float64)
for id, nc in enumerate(self.aens.ncl):
nc.setMolecule(coords=np.array(Xf, dtype=np.float32),
types=Sf)
E[id] = nc.energy()[0]
sig = np.std(hdn.hatokcal * E) / np.sqrt(Nai + Naj)
#print('Mol(',i,'): sig=',sig)
self.Nt += 1
if sig > sighat:
if sig > self.maxsig:
self.maxsig = sig
self.Nd += 1
hdn.writexyzfile(file + str(i).zfill(4) + '.xyz',
Xf.reshape(1, Xf.shape[0], 3), Sf)
self.frag_list.append(dict({'coords': Xf, 'spec': Sf}))
if sigma > 0.1:
Nb = Nb + 1
if key not in data.keys():
data[key] = (spc_l,[xyz_l])
else:
#print(type())
p_xyz = data[key][1]
p_xyz.append(xyz_l)
data[key] = (spc_l, p_xyz)
'''
print('H2O_frag' + str(i).zfill(3) + '.xyz : O=' + str(No).zfill(3) + ' : H=' + str(Nh).zfill(3) + ' : '
+ "{:13.3f}".format(energies[0]) +
' ' + "{:13.3f}".format(energies[1]) +
' ' + "{:13.3f}".format(energies[2]) +
' ' + "{:13.3f}".format(energies[3]) +
' ' + "{:13.3f}".format(energies[4]) +
' ' + "{:.3f}".format(sigma))
'''
else:
Ne = Ne + 1
print('Warning: fragmentation failed! No:',No,'Nh:',Nh)
print('Bad Frags:', Nb, 'of',Nf,'Error:',Ne,'Estd:',Etot/float(Nf), ' Avg. Frc. Std.: ', Ftot/float(Nf))
dyn.run(40) # Do 100 steps of MD
for key in data:
xyz = np.array(data[key][1])
spc = data[key][0]
print(key,xyz.shape[0])
hdt.writexyzfile(stdir + 'frag' + key + '.xyz', xyz, spc)
