def dist(X, xtitle, xlim, width):
"""
Plot histograms of data.
"""
fig, ax = plt.subplots(figsize=(8, 5))
if xlim is None:
xlim = (min(X)-2*width, max(X)+2*width)
X_bins = np.arange(xlim[0], xlim[1], width)
hist, bin_edges = np.histogram(a=X, bins=X_bins)
if xtitle == 'purchasability class':
align = 'center'
else:
align = 'edge'
ax.bar(
bin_edges[:-1],
hist,
align=align,
alpha=1.0,
width=width,
color='#2980B9',
edgecolor='k'
)
pfn.define_standard_plot(
ax,
xtitle=xtitle,
ytitle='count',
xlim=xlim,
ylim=None
)
return fig, ax
def cs_logPvsNHA(logPs, Xs, HlogPs, HXs):
fig, ax = plt.subplots(figsize=(8, 5))
xlim = (0, 40)
ylim = (-9, 14)
CS = [(1.0, 1.0, 1.0), (44 / 255, 62 / 255, 80 / 255)]
cm = colors.LinearSegmentedColormap.from_list('test', CS, N=10)
fig, ax, hist = pfn.twoD_histogram(X_data=Xs,
Y_data=logPs,
xlim=xlim,
ylim=ylim,
cmap=cm,
fig=fig,
ax=ax)
cbar = fig.colorbar(hist[3], ax=ax)
cbar.ax.set_ylabel('count', fontsize=16)
cbar.ax.tick_params(labelsize=16)
ax.scatter(HXs,
HlogPs,
c='#E74C3C',
edgecolors='k',
marker='o',
alpha=1.0,
s=120)
pfn.define_standard_plot(
ax,
ylim=ylim,
xlim=xlim,
# xtitle='number of heavy atoms',
ytitle=r'logP',
xtitle=r'no. heavy atoms',
)
fig.tight_layout()
fig.savefig(f'chemical_space_logPNHA.pdf', dpi=720, bbox_inches='tight')
def mid_plots(parameter_sets, molecules, test_mol, full_results, colours,
markers):
for t in parameter_sets:
fig, ax = plt.subplots()
for name in molecules:
if name not in test_mol:
continue
X = []
Y = []
Y_err = []
for i, v in enumerate(parameter_sets[t]):
RES = full_results[t][name][v]
_, _, mid_diam_avg, mid_diam_std, _ = RES
avg = float(mid_diam_avg)
std = float(mid_diam_std)
# if i == 0:
# ax.errorbar(float(v), avg, c=colours[name],
# yerr=std, fmt=markers[name], label=name)
# else:
# ax.errorbar(float(v), avg, c=colours[name],
# yerr=std, fmt=markers[name])
X.append(float(v))
Y.append(avg)
Y_err.append(std)
X = np.asarray(X)
Y = np.asarray(Y)
Y_err = np.asarray(Y_err)
ax.plot(X, Y, c=colours[name], marker=markers[name], label=name)
ax.fill_between(X,
Y - Y_err,
Y + Y_err,
alpha=0.2,
facecolor=colours[name])
if t == 'N_conformers':
t_lim = (0, 1100)
t_name = '$N$' # 'no. conformers'
if t == 'spacing':
t_lim = (0.2, 1.1)
t_name = r'grid spacing [$\mathrm{\AA}$]'
if t == 'vdw':
t_lim = (0.4, 1.1)
t_name = 'vdW scale parameter'
if t == 'boxMargin':
t_lim = (3, 9)
t_name = r'box margin [$\mathrm{\AA}$]'
pfn.define_standard_plot(
ax,
xtitle=t_name,
ytitle=r'avg. intermediate diameter [$\mathrm{\AA}$]',
xlim=t_lim,
ylim=(3.5, 9))
# ax.legend(fontsize=16, ncol=2)
fig.tight_layout()
fig.savefig(f"mid_{t}.pdf", dpi=720, bbox_inches='tight')
def cs_purchCT(purch, not_purch):
fig, ax = plt.subplots(figsize=(8, 5))
plot_prop = {
't': {
'c': '#FA7268',
'e': 'none',
'a': 0.5,
'm': 'o',
's': 50,
'label': 'purchasable'
},
'f': {
'c': '#DAF7A6',
'e': 'none',
'a': 0.5,
'm': 'x',
's': 50,
'label': 'not purchasable'
}
}
# bin each of the sets of data based on X value
for p in plot_prop:
pp = plot_prop[p]
if p == 't':
data = purch
else:
data = not_purch
width = 50
X_bins = np.arange(0, 2000, width)
hist, bin_edges = np.histogram(a=data, bins=X_bins, density=False)
ax.bar(
bin_edges[:-1],
hist,
align='edge',
alpha=0.8,
width=width,
color=pp['c'],
edgecolor='k',
label=pp['label'],
)
ax.legend(fontsize=16)
pfn.define_standard_plot(
ax,
# xtitle='number of heavy atoms',
xtitle=r'BertzCT',
ytitle='frequency',
)
fig.tight_layout()
fig.savefig(f'chemical_space_purchCT.pdf', dpi=720, bbox_inches='tight')
def shapes_with_known(molecules, known_df, threshold, output_dir):
"""
Plot molecule shapes considering experimental results.
"""
fig, ax = plt.subplots(figsize=(5, 5))
for name in molecules:
out_file = (f"{output_dir}/{name.replace(' ', '_').replace('/', '__')}"
'_diam_result.csv')
if os.path.exists(out_file) is False:
continue
results = pd.read_csv(out_file)
if len(results) == 0:
continue
mid_diam = min(results['diam2'])
lit_d = known_df[known_df['molecule'] == name]['diffuse'].iloc[0]
if lit_d == 't':
if mid_diam <= threshold:
C = 'b'
M = 'o'
else:
C = 'b'
M = 'X'
elif lit_d == 'f':
if mid_diam <= threshold:
C = 'r'
M = 'X'
else:
C = 'r'
M = 'o'
else:
continue
ax.scatter(np.average(results['ratio_1']),
np.average(results['ratio_2']),
c=C,
edgecolors='k',
marker=M,
alpha=1.0,
s=80)
ax.plot([0, 0.5, 1, 0], [1, 0.5, 1, 1], c='k', lw=2)
ax.text(0.75, 1.03, 'sphere', fontsize=20)
ax.text(0.4, 0.45, 'oblate', fontsize=20)
ax.text(-0.05, 1.03, 'prolate', fontsize=20)
pfn.define_standard_plot(ax,
xtitle='$I_1$ / $I_3$',
ytitle='$I_2$ / $I_3$',
xlim=(-0.1, 1.1),
ylim=(0.4, 1.1))
fig.tight_layout()
fig.savefig("shape.pdf", dpi=720, bbox_inches='tight')
def mol_parity(propx, propy, file, xtitle, ytitle, mol_file=None):
"""
Plot a parity of two molecular properties.
"""
if mol_file is None:
molecule_list = glob.glob('*_unopt.mol')
else:
molecule_list = IO.read_molecule_list(mol_file)
# iterate over molecules
Xs = []
Ys = []
for mol in molecule_list:
name = mol.replace('_unopt.mol', '')
prop_file = name + '_prop.json'
if not exists(prop_file):
continue
with open(prop_file, 'r') as f:
prop_dict = json.load(f)
Xs.append(prop_dict[propx])
Ys.append(prop_dict[propy])
fig, ax = plt.subplots(figsize=(8, 5))
ax.scatter(Xs,
Ys,
c='#FA7268',
edgecolors='k',
marker='o',
alpha=1.0,
s=80)
xlim = None
ylim = None
if propx == 'Synth_score':
xlim = (0, 10)
elif propy == 'Synth_score':
ylim = (0, 10)
pfn.define_standard_plot(ax,
xtitle=xtitle,
ytitle=ytitle,
xlim=xlim,
ylim=ylim)
fig.tight_layout()
fig.savefig(f'parity_{file}.pdf', dpi=720, bbox_inches='tight')
def biomin_known(molecules, output_dir, plot_suffix):
"""
Scatter plot of all molecule sizes in dictionary.
"""
fig, ax = plt.subplots(figsize=(8, 5))
m_diams = []
for name in molecules:
out_file = (f"{output_dir}/"
f"{name.replace(' ', '_').replace('/', '__')}"
'_diam_result.csv')
if os.path.exists(out_file) is False:
continue
results = pd.read_csv(out_file)
mid_diam = min(results['diam2'])
print('-----', name, mid_diam, '-----')
m_diams.append(mid_diam)
m_diams = np.asarray(m_diams)
X_bins = np.arange(0.1, 21, 0.5)
hist, bin_edges = np.histogram(a=m_diams, bins=X_bins)
ax.bar(bin_edges[:-1],
hist,
align='edge',
width=0.5,
color='#2C3E50',
edgecolor='k',
alpha=0.8)
ax.axvline(x=3.4, c='k')
ax.axvspan(
xmin=4.0,
xmax=6.6,
facecolor='k',
alpha=0.25,
# hatch="/"
)
# ax.axvspan(xmin=5.4, xmax=6.6, facecolor='k', alpha=0.2)
pfn.define_standard_plot(
ax,
# xtitle='intermediate diameter [$\mathrm{\AA}$]',
xtitle=r'$d$ [$\mathrm{\AA}$]',
ytitle='count',
xlim=(0, 15),
ylim=(0, 15))
fig.tight_layout()
fig.savefig(f"molecule_size_{plot_suffix}.pdf",
dpi=720,
bbox_inches='tight')
def min_of_mid_plots(parameter_sets, molecules, test_mol, full_results,
colours, markers):
for t in parameter_sets:
fig, ax = plt.subplots()
for name in molecules:
if name not in test_mol:
continue
X = []
Y = []
Y_err = []
for i, v in enumerate(parameter_sets[t]):
_, _, _, _, min_mid = full_results[t][name][v]
# if i == 0:
# ax.errorbar(float(v), avg, c=colours[name],
# yerr=std, fmt=markers[name], label=name)
# else:
# ax.errorbar(float(v), avg, c=colours[name],
# yerr=std, fmt=markers[name])
X.append(float(v))
Y.append(min_mid)
X = np.asarray(X)
Y = np.asarray(Y)
print(name, '--', max([i - min(Y) for i in Y]))
Y_err = np.asarray(Y_err)
ax.plot(X, Y, c=colours[name], marker=markers[name], label=name)
if t == 'N_conformers':
t_lim = (0, 1100)
t_name = '$N$' # 'no. conformers'
if t == 'spacing':
t_lim = (0.2, 0.7)
t_name = r'grid spacing [$\mathrm{\AA}$]'
if t == 'vdw':
t_lim = (0.4, 1.1)
t_name = 'vdW scale parameter'
if t == 'boxMargin':
t_lim = (3, 9)
t_name = r'box margin [$\mathrm{\AA}$]'
pfn.define_standard_plot(ax,
xtitle=t_name,
ytitle=r'$d$ [$\mathrm{\AA}$]',
xlim=t_lim,
ylim=(3.5, 8))
# ax.legend(fontsize=16, ncol=3)
fig.tight_layout()
fig.savefig(f"min_of_mid_{t}.pdf", dpi=720, bbox_inches='tight')
input(f'^^ is max dev from min_of_mid for {t}')
def cs_NHA(Xs, Ys):
fig, ax = plt.subplots(figsize=(8, 5))
ylim = (0, 17)
xlim = (0, 40)
CS = [(1.0, 1.0, 1.0), (44 / 255, 62 / 255, 80 / 255)]
cm = colors.LinearSegmentedColormap.from_list('test', CS, N=10)
fig, ax, hist = pfn.twoD_histogram(X_data=Xs,
Y_data=Ys,
xlim=xlim,
ylim=ylim,
cmap=cm,
fig=fig,
ax=ax)
cbar = fig.colorbar(hist[3], ax=ax)
cbar.ax.set_ylabel('count', fontsize=16)
cbar.ax.tick_params(labelsize=16)
#
# ax.scatter(
# Xs,
# Ys,
# c='#FF7900',
# edgecolors='k',
# marker='o',
# alpha=1.0,
# s=120
# )
# Horizontal lines for different materials.
ax.axhspan(ymin=4.0, ymax=6.6, facecolor='k', alpha=0.2)
# ax.axvspan(xmin=5.4, xmax=6.6, facecolor='k', alpha=0.2)
# plot possible region of ZIF pore limiting diameters from
# Banerjee 2008 - 10.1126/science.1152516
# ax.axvspan(0.0, 13, facecolor='#2ca02c', alpha=0.2)
# HOF size limit:
# ax.axvline(x=13.1, c='k', lw=2, linestyle='--')
pfn.define_standard_plot(
ax,
# xtitle='number of heavy atoms',
ylim=ylim,
xlim=xlim,
ytitle=r'intermediate diameter [$\mathrm{\AA}$]',
xtitle=r'no. heavy atoms',
)
fig.tight_layout()
fig.savefig(f'chemical_space_NHA.pdf', dpi=720, bbox_inches='tight')
def min_plots(parameter_sets, molecules, test_mol, full_results, colours,
markers):
for t in parameter_sets:
fig, ax = plt.subplots()
for name in molecules:
if name not in test_mol:
continue
X = []
Y = []
Y_err = []
for i, v in enumerate(parameter_sets[t]):
RES = full_results[t][name][v]
min_diam_avg, min_diam_std, _, _, _ = RES
avg = float(min_diam_avg)
std = float(min_diam_std)
# if i == 0:
# ax.errorbar(float(v), avg, c=colours[name],
# yerr=std, fmt=markers[name], label=name)
# else:
# ax.errorbar(float(v), avg, c=colours[name],
# yerr=std, fmt=markers[name])
X.append(float(v))
Y.append(avg)
Y_err.append(std)
ax.plot(X, Y, c=colours[name], marker=markers[name], label=name)
if t == 'N_conformers':
t_lim = (0, 1100)
t_name = '$N$' # 'no. conformers'
if t == 'spacing':
t_lim = (0, 1.2)
t_name = r'grid spacing [$\mathrm{\AA}$]'
if t == 'vdw':
t_lim = (0.4, 1.2)
t_name = r'vdW scale parameter'
if t == 'boxMargin':
t_lim = (2, 10)
t_name = r'box margin [$\mathrm{\AA}$]'
pfn.define_standard_plot(
ax,
xtitle=t_name,
ytitle=r'avg. minimum diameter [$\mathrm{\AA}$]',
xlim=t_lim,
ylim=(0, 10))
ax.legend(loc=1, fontsize=16)
fig.tight_layout()
fig.savefig(f"min_{t}.pdf", dpi=720, bbox_inches='tight')
def shapes(molecules, threshold, output_dir, plot_suffix):
"""
Plot molecule shapes of all molecules in dictionary.
"""
fig, ax = plt.subplots(figsize=(5, 5))
for name in molecules:
out_file = (f"{output_dir}/"
f"{name.replace(' ', '_').replace('/', '__')}"
'_diam_result.csv')
if os.path.exists(out_file) is False:
continue
results = pd.read_csv(out_file)
mid_diam = min(results['diam2'])
if mid_diam <= threshold:
C = 'b'
M = 'o'
E = 'k'
else:
C = 'r'
M = 'o'
E = 'k'
ax.scatter(np.average(results['ratio_1']),
np.average(results['ratio_2']),
c=C,
edgecolors=E,
marker=M,
alpha=1.0,
s=80)
ax.plot([0, 0.5, 1, 0], [1, 0.5, 1, 1], c='k', lw=2)
ax.text(0.75, 1.03, 'sphere', fontsize=20)
ax.text(0.4, 0.45, 'oblate', fontsize=20)
ax.text(-0.05, 1.03, 'prolate', fontsize=20)
pfn.define_standard_plot(ax,
title='',
xtitle='$I_1$ / $I_3$',
ytitle='$I_2$ / $I_3$',
xlim=(-0.1, 1.1),
ylim=(0.4, 1.1))
fig.tight_layout()
fig.savefig(f"shape_{plot_suffix}.pdf", dpi=720, bbox_inches='tight')
def parity_with_known_min2(molecules, diameters, output_dir):
"""
Parity plot of calculated diameters and known kinetic diameters.
"""
fig, ax = plt.subplots(figsize=(5, 5))
for name in molecules:
try:
min2_diam = float(diameters[name])
except ValueError:
print('no radius given for this molecule - skipped')
continue
out_file = (f"{output_dir}/{name.replace(' ', '_').replace('/', '__')}"
'_diam_result.csv')
if os.path.exists(out_file) is False:
continue
results = pd.read_csv(out_file)
if len(results) == 0:
continue
mid_diam = min(results['diam2'])
C = '#E74C3C'
M = 'o'
print(name, min2_diam, mid_diam)
ax.scatter(min2_diam,
mid_diam,
c=C,
edgecolors='k',
marker=M,
alpha=1.0,
s=120)
ax.plot(np.linspace(-1, 12, 2), np.linspace(-1, 12, 2), c='k', alpha=0.4)
# plot the limit from the two Sholl papers on diffusion
# ax.axvspan(4.0, 4.2, facecolor='r', alpha=0.5)
pfn.define_standard_plot(ax,
xtitle=r'critical diameter [$\mathrm{\AA}$]',
ytitle=r'$d$ [$\mathrm{\AA}$]',
xlim=(1, 10),
ylim=(1, 10))
fig.tight_layout()
fig.savefig("parity_min2.pdf", dpi=720, bbox_inches='tight')
def HOF_examples(output_dir):
"""
Prepare figure showing the value of d for all molecules used in the
BioHOFs from: 10.1021/jacs.9b06589
"""
# the n-phenyl esters
mol_list_1 = [
'fluorescein', 'hydrogen_peroxide', 'methanol', 'formaldehyde', 'urea'
]
smiles_list_1 = [
'C1=CC=C2C(=C1)C(=O)OC23C4=C(C=C(C=C4)O)OC5=C3C=CC(=C5)O', 'OO', 'CO',
'C=O', 'C(=O)(N)N'
]
fig, ax = plt.subplots(figsize=(8, 5))
for i, name in enumerate(mol_list_1):
out_file = (f"{output_dir}/"
f"{name.replace(' ', '_').replace('/', '__')}"
'_diam_result.csv')
if os.path.exists(out_file) is False:
continue
results = pd.read_csv(out_file)
mid_diam = min(results['diam2'])
mol = Chem.AddHs(Chem.MolFromSmiles(smiles_list_1[i]))
MW = Descriptors.MolWt(mol)
print(name, mol_list_1[i], MW, mid_diam)
ax.scatter(MW,
mid_diam,
c='#5499C7',
edgecolors='k',
marker='o',
alpha=1.0,
s=140)
# ax.axhline(y=11.8, c='k', alpha=0.2)
pfn.define_standard_plot(ax,
xtitle='molecular weight [g/mol]',
ytitle=r'$d$ [$\mathrm{\AA}$]',
xlim=(10, 500),
ylim=(2.5, 15))
fig.tight_layout()
fig.savefig("HOF_examples.pdf", dpi=720, bbox_inches='tight')
def cyt_C_perox_assay(output_dir):
"""
Prepare figure showing the change in intermediate diameter for 3
peroxide molcules degraded by Cyt-C in ZIF-8 (One-Pot Synthesis of
Protein-Embedded Metal–Organic Frameworks with Enhanced Biological
Activities, DOI:10.1021/nl5026419)
"""
# the n-phenyl esters
mol_list_1 = [
'hydrogen peroxide', 'methyl ethyl ketone peroxide',
'tert-butyl hydroperoxide'
]
smiles_list_1 = ['OO', 'CCC(C)(OO)OOC(C)(CC)OO', 'CC(C)(C)OO']
fig, ax = plt.subplots()
for i, name in enumerate(mol_list_1):
out_file = (f"{output_dir}/"
f"{name.replace(' ', '_').replace('/', '__')}"
'_diam_result.csv')
if os.path.exists(out_file) is False:
continue
results = pd.read_csv(out_file)
mid_diam = min(results['diam2'])
mol = Chem.AddHs(Chem.MolFromSmiles(smiles_list_1[i]))
MW = Descriptors.MolWt(mol)
print(name, mol_list_1[i], MW, mid_diam)
ax.scatter(MW,
mid_diam,
c='k',
edgecolors='k',
marker='o',
alpha=1.0,
s=100)
ax.axhspan(ymin=4.0, ymax=6.6, facecolor='k', alpha=0.2, hatch="/")
pfn.define_standard_plot(ax,
xtitle='molecular weight [g/mol]',
ytitle=r'$d$ [$\mathrm{\AA}$]',
xlim=(10, 250),
ylim=(2.5, 8))
fig.tight_layout()
fig.savefig("cytC_comp.pdf", dpi=720, bbox_inches='tight')
def mol_dist(data_dict):
"""
Plot distribution of a molecular property.
"""
fig, ax = plt.subplots(figsize=(8, 5))
width = data_dict['width']
X_bins = np.arange(data_dict['xlim'][0], data_dict['xlim'][1], width)
hist, bin_edges = np.histogram(a=data_dict['d'], bins=X_bins)
ax.bar(bin_edges[:-1],
hist,
align='edge',
alpha=1.0,
width=width,
color=data_dict['c'],
edgecolor='k')
pfn.define_standard_plot(ax,
xtitle=data_dict['xtitle'],
ytitle='count',
xlim=data_dict['xlim'],
ylim=None)
fig.tight_layout()
fig.savefig(f"hist_{data_dict['file']}.pdf", dpi=720, bbox_inches='tight')
def no_rxns_vs_size(data, params, plot_suffix):
"""
Plot number of possible reactions as a function of size threshold.
"""
fig, ax = plt.subplots(figsize=(8, 5))
# bin each of the sets of data based on X value
width = 0.5
X_bins = np.arange(0, 20.5, width)
hist, bin_edges = np.histogram(a=data['max_mid_diam'], bins=X_bins)
ax2 = ax.twinx()
ax2.bar(
bin_edges[:-1],
hist,
align='edge',
alpha=0.9, width=width,
color='#2C3E50',
edgecolor='k'
)
# cumulative plot
cumul = np.cumsum(hist)
ax.plot(
bin_edges[:-1],
cumul,
alpha=1.0,
label='max component < threshold',
color='r',
marker='o'
)
# ax.axvspan(xmin=4.0, xmax=6.6, facecolor='k', alpha=0.2,
# hatch="/")
ax.axvspan(xmin=4.0, xmax=6.6, facecolor='k', alpha=0.2)
# ax.axvspan(xmin=5.4, xmax=6.6, facecolor='k', alpha=0.2)
# plot possible region of ZIF pore limiting diameters from
# Banerjee 2008 - 10.1126/science.1152516
# ax.axvspan(0.0, 13, facecolor='#2ca02c', alpha=0.2)
# ax.axvline(x=13.1, c='k', lw=2, linestyle='--')
pfn.define_standard_plot(
ax,
xtitle=r'$d$ of largest component [$\mathrm{\AA}$]',
ytitle='cumulative # reactions',
xlim=(0, 17),
ylim=(0, int(max(cumul)+max(cumul)*0.1))
)
ax2.set_ylim(0, int(max(hist)+max(hist)*0.2))
ax2.set_ylabel('# reactions', fontsize=16)
ax.yaxis.set_major_locator(MaxNLocator(integer=True))
ax2.yaxis.set_major_locator(MaxNLocator(integer=True))
# Change left y axis colours.
ax.spines['left'].set_color('red')
ax2.spines['left'].set_color('red')
ax2.tick_params(axis='both', which='major', labelsize=16)
fig.tight_layout()
fig.savefig(
f"{plot_suffix}/size_threshold_{plot_suffix}.pdf",
dpi=720,
bbox_inches='tight'
)
def n_phenyl_assay(output_dir):
"""
Prepare figure showing the change in intermediate diameter for
molecules
commonly used in n-phenyl ester hydrolysis assays.
"""
# the n-phenyl esters
mol_list_1 = [
'p-nitrophenyl acetate', 'p-nitrophenyl butyrate',
'p-nitrophenyl hexanoate', 'p-nitrophenyl octanoate',
'p-nitrophenyl decanoate', 'p-nitrophenyl dodecanoate'
]
# the products
mol_list_2 = [
'acetic acid', 'butyric acid', 'hexanoic acid', 'octanoic acid',
'decanoic acid', 'dodecanoic acid'
]
# no Cs
no_Cs = [2, 4, 6, 8, 10, 12]
fig, ax = plt.subplots()
for i, C in enumerate(no_Cs):
# ester
name = mol_list_1[i]
out_file = (f"{output_dir}/"
f"{name.replace(' ', '_').replace('/', '__')}"
'_diam_result.csv')
if os.path.exists(out_file) is False:
continue
results = pd.read_csv(out_file)
mid_diam = min(results['diam2'])
print(C, mol_list_1[i], mid_diam)
ax.scatter(C,
mid_diam,
c='r',
edgecolors='k',
marker='o',
alpha=1.0,
s=100)
# acid
name = mol_list_2[i]
out_file = (f"{output_dir}/"
f"{name.replace(' ', '_').replace('/', '__')}"
'_diam_result.csv')
if os.path.exists(out_file) is False:
continue
results = pd.read_csv(out_file)
mid_diam = min(results['diam2'])
print(C, mol_list_2[i], mid_diam)
ax.scatter(C,
mid_diam,
c='b',
edgecolors='k',
marker='o',
alpha=1.0,
s=120)
ax.axhspan(ymin=4.0, ymax=6.6, facecolor='k', alpha=0.2, hatch="/")
# n-phenol
name = 'p-nitrophenol'
out_file = (f"{output_dir}/"
f"{name.replace(' ', '_').replace('/', '__')}"
'_diam_result.csv')
if os.path.exists(out_file) is False:
import sys
sys.exit('calc molecule diameters!')
results = pd.read_csv(out_file)
mid_diam = min(results['diam2'])
ax.axhline(y=mid_diam, c='purple', alpha=1)
pfn.define_standard_plot(ax,
xtitle='no. carbons',
ytitle=r'$d$ [$\mathrm{\AA}$]',
xlim=(1, 14),
ylim=(2.5, 8))
# decoy legend
ax.scatter(-100,
-100,
c='r',
edgecolors='k',
marker='o',
alpha=1.0,
s=100,
label='ester')
ax.scatter(-100,
-100,
c='b',
edgecolors='k',
marker='o',
alpha=1.0,
s=100,
label='acid')
ax.legend(fontsize=16)
fig.tight_layout()
fig.savefig("ester_comp.pdf", dpi=720, bbox_inches='tight')
def rxn_space(data, filename):
"""
Plot number of possible reactions as a function of size threshold.
"""
plot_prop = {
1: {
'c': '#FA7268',
'e': 'none',
'a': 0.5,
'm': 'o',
's': 50,
'label': 'class I'
},
2: {
'c': '#DAF7A6',
'e': 'none',
'a': 0.5,
'm': 'x',
's': 50,
'label': 'class II'
},
3: {
'c': '#900C3F',
'e': 'none',
'a': 1.0,
'm': 'x',
's': 50,
'label': 'class III'
},
4: {
'c': '#F6D973',
'e': 'none',
'a': 0.5,
'm': 'x',
's': 50,
'label': 'class IV'
}
}
# bin each of the sets of data based on X value
width = 0.5
X_bins = np.arange(0, 20.5, width)
fig, ax = plt.subplots(figsize=(8, 5))
# bin each of the sets of data based on X value
for p in plot_prop:
if p != 3:
continue
pp = plot_prop[p]
sub_data = data[data['PC_class'] == p]
hist, bin_edges = np.histogram(
a=sub_data['max_mid_diam'],
bins=X_bins
)
ax.bar(
bin_edges[:-1],
hist,
align='edge',
alpha=pp['a'],
width=width,
color=pp['c'],
edgecolor='k',
label=pp['label']
)
ax.legend(fontsize=16)
ax.axvspan(xmin=4.0, xmax=6.6, facecolor='k', alpha=0.2, hatch="/")
# ax.axvspan(xmin=5.4, xmax=6.6, facecolor='k', alpha=0.2)
# plot possible region of ZIF pore limiting diameters from
# Banerjee 2008 - 10.1126/science.1152516
# ax.axvspan(0.0, 13, facecolor='#2ca02c', alpha=0.2)
# HOF.
ax.axvline(x=13.1, c='k', lw=2, linestyle='--')
pfn.define_standard_plot(
ax,
xtitle=r'$d$ of largest component [$\mathrm{\AA}$]',
ytitle='# reactions',
xlim=(0, 17),
ylim=None
)
fig.tight_layout()
fig.savefig(
filename,
dpi=720,
bbox_inches='tight'
)
def cs_purch(purch, not_purch):
fig, ax = plt.subplots(figsize=(8, 5))
plot_prop = {
't': {
'c': '#FA7268',
'e': 'none',
'a': 0.5,
'm': 'o',
's': 50,
'label': 'purchasable'
},
'f': {
'c': '#DAF7A6',
'e': 'none',
'a': 0.5,
'm': 'x',
's': 50,
'label': 'not purchasable'
}
}
# bin each of the sets of data based on X value
for p in plot_prop:
pp = plot_prop[p]
if p == 't':
data = purch
else:
data = not_purch
width = 0.5
X_bins = np.arange(0, 15.5, width)
hist, bin_edges = np.histogram(a=data, bins=X_bins, density=True)
ax.bar(
bin_edges[:-1],
hist,
align='edge',
alpha=0.8,
width=width,
color=pp['c'],
edgecolor='k',
label=pp['label'],
)
# for X, Y, Z in zip(Xs, Ys, Zs):
# if Z:
# pp = plot_prop['t']
# else:
# pp = plot_prop['f']
#
# ax.scatter(
# X,
# Y,
# c=pp['c'],
# edgecolors=pp['e'],
# marker=pp['m'],
# alpha=pp['a'],
# s=pp['s']
# )
# Vertical lines for different materials.
ax.axvspan(xmin=4.0, xmax=6.6, facecolor='k', alpha=0.2, hatch="/")
# ax.axvspan(xmin=5.4, xmax=6.6, facecolor='k', alpha=0.2)
# plot possible region of ZIF pore limiting diameters from
# Banerjee 2008 - 10.1126/science.1152516
# ax.axvspan(0.0, 13, facecolor='#2ca02c', alpha=0.2)
# HOF size limit:
ax.axvline(x=13.1, c='k', lw=2, linestyle='--')
# # Legend.
# for p in plot_prop:
# pp = plot_prop[p]
# ax.scatter(
# X,
# Y,
# c=pp['c'],
# edgecolors=pp['e'],
# marker=pp['m'],
# alpha=pp['a'],
# s=pp['s'],
# label=pp['label']
# )
ax.legend(fontsize=16)
pfn.define_standard_plot(
ax,
# xtitle='number of heavy atoms',
xtitle=r'intermediate diameter [$\mathrm{\AA}$]',
ytitle='frequency',
)
fig.tight_layout()
fig.savefig(f'chemical_space_purch.pdf', dpi=720, bbox_inches='tight')
def parity_cf_scale_with_known(molecules, diameters, known_df, pars,
scale_info):
"""
Produce a parity plot of calculated diameters and known kinetic
diameters for multiple input parameters.
"""
S = 120
fig, ax = plt.subplots(figsize=(5, 5))
for dir in scale_info:
if dir != 'scale09_test':
continue
kin_diams = []
mid_diams = []
sc, C, M, A, E = scale_info[dir]
scale_output = f'scale_sc_{dir}.txt'
if os.path.exists(scale_output):
with open(scale_output, 'r') as f:
for line in f:
res = line.rstrip().split('__')
name, kin_diam, mid_diam = res
kin_diams.append(float(kin_diam))
mid_diams.append(float(mid_diam))
ax.scatter(float(kin_diam),
float(mid_diam),
c=C,
edgecolors=E,
marker=M,
alpha=A,
s=S)
else:
with open(scale_output, 'w') as f:
for name in molecules:
try:
kin_diam = float(diameters[name])
except ValueError:
print('no radius given for this molecule ' '- skipped')
continue
out_file = (f"{dir}/"
f"{name.replace(' ', '_').replace('/', '__')}"
'_diam_result.csv')
if os.path.exists(out_file) is False:
continue
results = pd.read_csv(out_file)
if len(results) == 0:
continue
mid_diam = min(results['diam2'])
kin_diams.append(float(kin_diam))
mid_diams.append(float(mid_diam))
ax.scatter(float(kin_diam),
float(mid_diam),
c=C,
edgecolors=E,
marker=M,
alpha=A,
s=S)
f.write(name + '__' + str(kin_diam) + '__' +
str(mid_diam) + '\n')
corr = pearsonr(kin_diams, mid_diams)
MAE = mean_absolute_error(kin_diams, mid_diams)
chi2 = sum([((j - i)**2) / i for i, j in zip(kin_diams, mid_diams)])
print(f'{dir} R^2: {corr}, MAE: {MAE}, chi^2: {chi2}')
ax.plot(np.linspace(-1, 12, 2), np.linspace(-1, 12, 2), c='k', alpha=0.4)
pfn.define_standard_plot(
ax,
xtitle=r'kinetic diameter [$\mathrm{\AA}$]',
# ytitle='intermediate diameter [$\mathrm{\AA}$]',
ytitle=r'$d$ [$\mathrm{\AA}$]',
xlim=(1, 10),
ylim=(1, 10))
# legend
for dir in scale_info:
if dir != 'scale09_test':
continue
sc, C, M, A, E = scale_info[dir]
ax.scatter(-100,
-100,
c=C,
edgecolors=E,
marker=M,
alpha=A,
s=S,
label=f'vdW scale = {sc}')
# ax.legend(loc=2, fontsize=14)
fig.tight_layout()
fig.savefig("parity_scalecf.pdf", dpi=720, bbox_inches='tight')
def dist_cf_scale_with_known(molecules, diameters, known_df, pars, scale_info):
"""
Produce a bar plot of distributions of the deviations of calculated
diameters and known kinetic diameters for multiple input params.
"""
fig, ax = plt.subplots(figsize=(8, 5))
for dir in scale_info:
kin_diams = []
mid_diams = []
sc, C, M, A, E = scale_info[dir]
scale_output = f'scale_sc_{dir}.txt'
if os.path.exists(scale_output):
with open(scale_output, 'r') as f:
for line in f:
res = line.rstrip().split('__')
name, kin_diam, mid_diam = res
kin_diams.append(float(kin_diam))
mid_diams.append(float(mid_diam))
else:
with open(scale_output, 'w') as f:
for name in molecules:
try:
kin_diam = float(diameters[name])
except ValueError:
print('no radius given for this molecule ' '- skipped')
continue
out_file = (f"{dir}/"
f"{name.replace(' ', '_').replace('/', '__')}"
'_diam_result.csv')
if os.path.exists(out_file) is False:
continue
results = pd.read_csv(out_file)
if len(results) == 0:
continue
mid_diam = min(results['diam2'])
kin_diams.append(float(kin_diam))
mid_diams.append(float(mid_diam))
f.write(name + '__' + str(kin_diam) + '__' +
str(mid_diam) + '\n')
corr = pearsonr(kin_diams, mid_diams)
MAE = mean_absolute_error(kin_diams, mid_diams)
chi2 = sum([((j - i)**2) / i for i, j in zip(kin_diams, mid_diams)])
print(f'{dir} R^2: {corr}, MAE: {MAE}, chi^2: {chi2}')
X_vals = [i - j for i, j in zip(mid_diams, kin_diams)]
width = 0.1
xlim = (-2, 2)
X_bins = np.arange(xlim[0], xlim[1], width)
hist, bin_edges = np.histogram(a=X_vals, bins=X_bins)
#
# ax.bar(
# bin_edges[:-1],
# hist,
# align='edge',
# alpha=0.5,
# width=width,
# color=C,
# edgecolor='k',
# label=f'vdW scale = {sc}'
# )
ax.plot(X_bins[:-1] + width / 2,
hist,
c=C,
lw='1.5',
marker='o',
alpha=1.0,
label=f'vdW scale = {sc}')
pfn.define_standard_plot(
ax,
xtitle=r'|$d$ - kinetic diameter| [$\mathrm{\AA}$]',
ytitle='count',
xlim=xlim,
ylim=None)
ax.legend(fontsize=14)
fig.tight_layout()
fig.savefig("dist_scalecf.pdf", dpi=720, bbox_inches='tight')
def target_conformer_plot(parameter_sets, molecules, test_mol, full_results,
colours, markers, properties):
# target no conformers
targ_confs = [50, 200]
# set property
for p, PROP in enumerate(['MW', 'NHA', 'NRB']):
if p == 0:
PROP_lab = 'MW [g/mol]'
p_lim = (0, 120)
if p == 1:
PROP_lab = 'no. heavy atoms'
p_lim = (0, 9)
if p == 2:
PROP_lab = 'no. rotatable bonds'
p_lim = (0, 6)
for t in parameter_sets:
if t != 'N_conformers':
continue
# fig = plt.figure() # figsize=(8, 8))
# ax = fig.add_subplot(111, projection='3d')
fig, ax = plt.subplots()
for name in molecules:
if name not in test_mol:
continue
X = []
Y = []
Z = []
for i, v in enumerate(parameter_sets[t]):
RES = full_results[t][name][v]
_, _, _, _, min_mid = RES
# if i == 0:
# ax.errorbar(
# float(v), avg, c=colours[name],
# yerr=std, fmt=markers[name], label=name)
# else:
# ax.errorbar(float(v), avg, c=colours[name],
# yerr=std, fmt=markers[name])
X.append(float(v))
Y.append(min_mid)
Z.append(properties[name][p])
X = np.asarray(X)
Y = np.asarray(Y)
Z = np.asarray(Z)
for targ_conf in targ_confs:
Y2 = Y - Y[-1]
Z2 = Z[X == targ_conf]
Y2 = Y2[X == targ_conf]
# plot points
# ax.scatter(X, Y-Y[-1], Z, s=60,
# c=colours[name], marker=markers[name])
ax.scatter(Z2,
Y2,
c=colours[name],
marker=markers[name],
label=name,
s=80)
pfn.define_standard_plot(
ax,
xtitle=PROP_lab,
# ytitle=(
# '$d_{\mathrm{i, min}}$ - '
# '$d_{\mathrm{i, min}}$(1000) '
# '[$\mathrm{\AA}$]'
# ),
ytitle=r'$d-d$(1000) [$\mathrm{\AA}$]',
xlim=p_lim,
ylim=(-0.1, 0.5))
ax.axhline(y=0, c='k', linestyle='--')
# ax.set_xlabel(t_name, fontsize=16)
# ax.set_ylabel(
# '$d_{\mathrm{i, min}}-d_{\mathrm{i, min}}$(1000)'
# ' [$\mathrm{\AA}$]'
# ),
# fontsize=16)
# ax.set_zlabel(PROP_lab, fontsize=16)
# ax.set_xlim(t_lim)
# ax.set_ylim(-0.1, 0.5)
# ax.set_zlim(p_lim)
# ax.set_aspect('equal', 'box')
# dist = 30
# angles = 10
# ax.view_init(dist, angles)
# ax.legend(fontsize=14, ncol=2)
fig.tight_layout()
fig.savefig(f"min_of_mid_{t}_v_prop_{PROP}.pdf",
bbox_inches='tight',
dpi=720)
def seed_test(seeds):
"""
Compares the minimum diameter obtained for a set of molecules with
different random seeds for the ETKDG algorithm.
"""
molecules = {
'n-hexane': 'CCCCCC',
'n-heptane': 'CCCCCCC',
'n-octane': 'CCCCCCCC',
'toluene': 'CC1=CC=CC=C1',
'p-nitrophenol': 'C1=CC(=CC=C1[N+](=O)[O-])O',
'p-nitrophenyl butyrate': 'CCCC(=O)OC1=CC=C(C=C1)[N+](=O)[O-]',
'butyric acid': 'CCCC(=O)O',
}
colours = {
'n-hexane': 'k',
'n-heptane': 'r',
'n-octane': 'b',
'toluene': 'green',
'p-nitrophenol': 'purple',
'p-nitrophenyl butyrate': 'orange',
'butyric acid': 'darkgray',
}
markers = {
'n-hexane': 'o',
'n-heptane': 'X',
'n-octane': 'D',
'toluene': 'P',
'p-nitrophenol': '^',
'p-nitrophenyl butyrate': '>',
'butyric acid': '<',
}
seed_output = "seed_test.pkl"
if os.path.exists(seed_output):
# load results
full_results = pickle.load(open(seed_output, "rb"))
else:
full_results = {}
for t in seeds:
full_results[t] = {}
for name in molecules:
full_results[t][name] = {}
for name in molecules:
for t in seeds:
output_dir = f'seeds_{t}'
out_file = (f"{output_dir}/"
f"{name.replace(' ', '_').replace('/', '__')}"
'_diam_result.csv')
if os.path.exists(out_file) is False:
continue
results = pd.read_csv(out_file)
if len(results) == 0:
continue
min_diam_avg = np.average(results['diam1'])
min_diam_std = np.std(results['diam1'])
mid_diam_avg = np.average(results['diam2'])
mid_diam_std = np.std(results['diam2'])
min_mid = min(results['diam2'])
result = (min_diam_avg, min_diam_std, mid_diam_avg,
mid_diam_std, min_mid)
full_results[t][name] = result
# save file
pickle.dump(full_results, open("seed_test.pkl", "wb"))
fig, ax = plt.subplots()
for name in molecules:
X = []
Y = []
for t in seeds:
RES = full_results[t][name]
_, _, _, _, min_mid = RES
X.append(int(t))
Y.append(min_mid)
print(name, '--', max([i - min(Y) for i in Y]))
ax.scatter(X, Y, c=colours[name], marker=markers[name], label=name)
input('^^ max dev from min_of_mid for seeds')
t_lim = (0, 850000)
t_name = 'random seed'
pfn.define_standard_plot(ax,
xtitle=t_name,
ytitle=r'$d$ [$\mathrm{\AA}$]',
xlim=t_lim,
ylim=(4, 8))
# ax.set_xticks([1, 2, 3, 4, 5, 6, 7])
# ax.set_xticklabels([str(i) for i in seeds])
# ax.legend(fontsize=16, ncol=3)
fig.tight_layout()
fig.savefig("min_of_mid_seeds.pdf", dpi=720, bbox_inches='tight')
def rxn_complexity(data, filename):
"""
Plot the measures of complexity of each reaction.
"""
fig, ax = plt.subplots(figsize=(8, 5))
ylim = (-1000, 1000)
xlim = (-10, 10)
# CS = [(1.0, 1.0, 1.0), (44/255, 62/255, 80/255)]
# cm = colors.LinearSegmentedColormap.from_list('test', CS, N=10)
# fig, ax, hist = pfn.twoD_histogram(
# X_data=data['deltasa'],
# Y_data=data['deltabct'],
# xlim=xlim,
# ylim=ylim,
# cmap=cm,
# fig=fig,
# ax=ax
# )
# cbar = fig.colorbar(hist[3], ax=ax)
# cbar.ax.set_ylabel('count', fontsize=16)
# cbar.ax.tick_params(labelsize=16)
ax.scatter(
data['deltasa'],
data['deltabct'],
c='#CCD1D1',
edgecolors='none',
marker='o',
alpha=1.0,
s=40,
label='full dataset'
)
small_data = data[data['max_mid_diam'] < 6.6]
ax.scatter(
small_data['deltasa'],
small_data['deltabct'],
c='#2C3E50',
edgecolors='none',
marker='o',
alpha=1.0,
s=40,
label='viable reactions'
)
pfn.define_standard_plot(
ax,
# xtitle='number of heavy atoms',
ylim=ylim,
xlim=xlim,
ytitle=r'$\Delta$ BertzCT',
xtitle=r'$\Delta$ SAscore',
)
ax.legend(fontsize=16)
fig.tight_layout()
fig.savefig(
filename,
dpi=720,
bbox_inches='tight'
)
