def main():
# change this to False to produce the equation dataset
molecules = True
# change this to False to get character-based encodings instead of grammar-based
grammar = 'new' #use True for the grammar used by Kusner et al
# can't define model class inside settings as it itself uses settings a lot
_, my_model = get_vae(molecules, grammar)
def pre_parser(x):
try:
return next(my_model._parser.parse(x))
except Exception as e:
return None
settings = get_settings(molecules,grammar)
MAX_LEN = settings['max_seq_length']
#feature_len = settings['feature_len']
dest_file = settings['data_path']
source_file = settings['source_data']
# Read in the strings
f = open(source_file,'r')
L = []
for line in f:
line = line.strip()
L.append(line)
f.close()
# convert to one-hot and save, in small increments to save RAM
#dest_file = dest_file.replace('.h5','_new.h5')
ds = IncrementingHDF5Dataset(dest_file)
step = 100
dt = h5py.special_dtype(vlen=str) # PY3 hdf5 datatype for variable-length Unicode strings
size = min(10000, len(L))
for i in tqdm(range(0, size, step)):#for i in range(0, 1000, 2000):
#print('Processing: i=[' + str(i) + ':' + str(i + step) + ']')
these_indices = list(range(i, min(i + step,len(L))))
these_smiles = L[i:min(i + step,len(L))]
if grammar=='new': # have to weed out non-parseable strings
tokens = [my_model._tokenize(s.replace('-c','c')) for s in these_smiles]
these_smiles, these_indices = list(zip(*[(s,ind) for s,t,ind in zip(these_smiles, tokens, these_indices) if pre_parser(t) is not None]))
#print(len(these_smiles))
these_actions = torch.tensor(my_model.strings_to_actions(these_smiles))
action_seq_length = my_model.action_seq_length(these_actions)
onehot = my_model.actions_to_one_hot(these_actions)
append_data = {'smiles': np.array(these_smiles, dtype=dt),
'indices': np.array(these_indices),
'actions': these_actions,
'valid': np.ones((len(these_smiles))),
'seq_len': action_seq_length,
'data': onehot}
if molecules:
from rdkit.Chem.rdmolfiles import MolFromSmiles
mols = [MolFromSmiles(s) for s in these_smiles]
raw_scores = np.array([get_score_components_from_mol(m) for m in mols])
append_data['raw_scores'] = raw_scores
num_atoms = np.array([len(m.GetAtoms()) for m in mols])
append_data['num_atoms'] = num_atoms
ds.append(append_data)
if molecules:
# also calculate mean and std of the scores, to use in the ultimate objective
raw_scores = np.array(ds.h5f['raw_scores'])
score_std = raw_scores.std(0)
score_mean = raw_scores.mean(0)
ds.append_to_dataset('score_std',score_std)
ds.append_to_dataset('score_mean', score_mean)
print('success!')
grammar=True,
BATCH_SIZE=150,
drop_rate=0.3,
sample_z=True,
save_file='next_gen.h5',
encoder_type=False,
lr=5e-4,
plot_prefix='RNN enc lr 1e-4',
dashboard=dash_name,
preload_weights=False)
# this is a wrapper for encoding/decodng
grammar_model = ZincGrammarModel(model=model)
validity_model = to_gpu(
DenseHead(model.encoder, body_out_dim=settings['z_size'], drop_rate=0.3))
valid_smile_ds = IncrementingHDF5Dataset('valid_smiles.h5', valid_frac=0.1)
invalid_smile_ds = IncrementingHDF5Dataset('invalid_smiles.h5', valid_frac=0.1)
valid_fitter = train_validity(grammar=grammar,
model=validity_model,
EPOCHS=100,
BATCH_SIZE=40,
lr=1e-4,
main_dataset=main_dataset,
new_datasets=(valid_smile_ds, invalid_smile_ds),
save_file=None,
plot_prefix='valid_model',
dashboard=dash_name,
preload_weights=False)
# TODO: collect the smiles strings too, just for kicks, into hdf5!
from generative_playground.molecules.rdkit_utils.rdkit_utils import fraction_valid
import numpy as np
molecules = True
grammar = True
settings = get_settings(molecules, grammar)
dash_name = 'test'
visdom = Dashboard(dash_name)
model, grammar_model = get_vae(molecules,
grammar,
drop_rate=0.5,
decoder_type='action') # or 'action','old','step','attention'
reinforcement_model = ReinforcementModel(model.decoder)
h5_prefix = 'new4_'
valid_smile_ds = IncrementingHDF5Dataset(h5_prefix +'valid_smiles.h5')
invalid_smile_ds = IncrementingHDF5Dataset(h5_prefix + 'invalid_smiles.h5')
original_ds = IncrementingHDF5Dataset('../data/zinc_grammar_dataset.h5', mode='r')
RL_fitter = train_reinforcement(grammar = grammar,
model = reinforcement_model,
EPOCHS = 10000,
BATCH_SIZE = 25,
lr = 1e-4,
new_datasets = (valid_smile_ds, invalid_smile_ds, original_ds),
save_file = 'first_reinforcement.h5',
plot_prefix = 'valid_model',
dashboard = dash_name,
preload_weights=False)
count = 0
def train_policy_gradient(molecules=True,
grammar=True,
EPOCHS=None,
BATCH_SIZE=None,
reward_fun_on=None,
reward_fun_off=None,
max_steps=277,
lr_on=2e-4,
lr_off=1e-4,
drop_rate=0.0,
plot_ignore_initial=0,
save_file=None,
preload_file=None,
anchor_file=None,
anchor_weight=0.0,
decoder_type='action',
plot_prefix='',
dashboard='policy gradient',
smiles_save_file=None,
on_policy_loss_type='best',
off_policy_loss_type='mean',
sanity_checks=True):
root_location = os.path.dirname(
os.path.abspath(inspect.getfile(inspect.currentframe())))
root_location = root_location + '/../'
save_path = root_location + 'pretrained/' + save_file
smiles_save_path = root_location + 'pretrained/' + smiles_save_file
settings = get_settings(molecules=molecules, grammar=grammar)
if EPOCHS is not None:
settings['EPOCHS'] = EPOCHS
if BATCH_SIZE is not None:
settings['BATCH_SIZE'] = BATCH_SIZE
save_dataset = IncrementingHDF5Dataset(smiles_save_path)
task = SequenceGenerationTask(molecules=molecules,
grammar=grammar,
reward_fun=reward_fun_on,
batch_size=BATCH_SIZE,
max_steps=max_steps,
save_dataset=save_dataset)
def get_model(sanity_checks=sanity_checks):
return get_decoder(
molecules,
grammar,
z_size=settings['z_size'],
decoder_hidden_n=200,
feature_len=None, # made redundant, need to factor it out
max_seq_length=max_steps,
drop_rate=drop_rate,
decoder_type=decoder_type,
task=task)[0]
model = get_model()
if preload_file is not None:
try:
preload_path = root_location + 'pretrained/' + preload_file
model.load_state_dict(torch.load(preload_path))
except:
pass
anchor_model = None
if anchor_file is not None:
anchor_model = get_model()
try:
anchor_path = root_location + 'pretrained/' + anchor_file
anchor_model.load_state_dict(torch.load(anchor_path))
except:
anchor_model = None
from generative_playground.molecules.rdkit_utils.rdkit_utils import NormalizedScorer
import rdkit.Chem.rdMolDescriptors as desc
import numpy as np
scorer = NormalizedScorer()
def model_process_fun(model_out, visdom, n):
from rdkit import Chem
from rdkit.Chem.Draw import MolToFile
actions, logits, rewards, terminals, info = model_out
smiles, valid = info
total_rewards = rewards.sum(1)
best_ind = torch.argmax(total_rewards).data.item()
this_smile = smiles[best_ind]
mol = Chem.MolFromSmiles(this_smile)
pic_save_path = root_location + 'images/' + 'test.svg'
if mol is not None:
try:
MolToFile(mol, pic_save_path, imageType='svg')
with open(pic_save_path, 'r') as myfile:
data = myfile.read()
data = data.replace('svg:', '')
visdom.append('best molecule of batch', 'svg', svgstr=data)
except:
pass
scores, norm_scores = scorer.get_scores([this_smile])
visdom.append(
'score component',
'line',
X=np.array([n]),
Y=np.array(
[[x for x in norm_scores[0]] + [norm_scores[0].sum()] +
[scores[0].sum()] + [desc.CalcNumAromaticRings(mol)]]),
opts={
'legend': [
'logP', 'SA', 'cycle', 'norm_reward', 'reward',
'Aromatic rings'
]
})
visdom.append('fraction valid',
'line',
X=np.array([n]),
Y=np.array([valid.mean().data.item()]))
if reward_fun_off is None:
reward_fun_off = reward_fun_on
def get_fitter(model,
loss_obj,
fit_plot_prefix='',
model_process_fun=None,
lr=None,
loss_display_cap=float('inf'),
anchor_model=None,
anchor_weight=0):
nice_params = filter(lambda p: p.requires_grad, model.parameters())
optimizer = optim.Adam(nice_params, lr=lr)
scheduler = lr_scheduler.StepLR(optimizer, step_size=100, gamma=0.99)
if dashboard is not None:
metric_monitor = MetricPlotter(
plot_prefix=fit_plot_prefix,
loss_display_cap=loss_display_cap,
dashboard_name=dashboard,
plot_ignore_initial=plot_ignore_initial,
process_model_fun=model_process_fun)
else:
metric_monitor = None
checkpointer = Checkpointer(valid_batches_to_checkpoint=1,
save_path=save_path,
save_always=True)
def my_gen():
for _ in range(1000):
yield to_gpu(torch.zeros(BATCH_SIZE, settings['z_size']))
fitter = fit_rl(train_gen=my_gen,
model=model,
optimizer=optimizer,
scheduler=scheduler,
epochs=EPOCHS,
loss_fn=loss_obj,
grad_clip=5,
anchor_model=anchor_model,
anchor_weight=anchor_weight,
metric_monitor=metric_monitor,
checkpointer=checkpointer)
return fitter
# the on-policy fitter
fitter1 = get_fitter(model,
PolicyGradientLoss(on_policy_loss_type),
plot_prefix + 'on-policy',
model_process_fun=model_process_fun,
lr=lr_on,
anchor_model=anchor_model,
anchor_weight=anchor_weight)
# get existing molecule data to add training
main_dataset = DatasetFromHDF5(settings['data_path'], 'actions')
# TODO change call to a simple DataLoader, no validation
train_loader, valid_loader = train_valid_loaders(main_dataset,
valid_fraction=0.1,
batch_size=BATCH_SIZE,
pin_memory=use_gpu)
fitter2 = get_fitter(model,
PolicyGradientLoss(off_policy_loss_type),
plot_prefix + ' off-policy',
lr=lr_off,
model_process_fun=model_process_fun,
loss_display_cap=125)
def on_policy_gen(fitter, model):
while True:
model.policy = SoftmaxRandomSamplePolicy()
yield next(fitter)
def off_policy_gen(fitter, data_gen, model):
while True:
data_iter = data_gen.__iter__()
try:
x_actions = next(data_iter).to(torch.int64)
model.policy = PolicyFromTarget(x_actions)
yield next(fitter)
except StopIteration:
data_iter = data_gen.__iter__()
return model, on_policy_gen(fitter1,
model), off_policy_gen(fitter2, train_loader,
model)
MAX_LEN = settings['max_seq_length']
feature_len = settings['feature_len']
dest_file = settings['data_path']
source_file = settings['source_data']
# Read in the strings
f = open(source_file, 'r')
L = []
for line in f:
line = line.strip()
L.append(line)
f.close()
# convert to one-hot and save, in small increments to save RAM
#dest_file = dest_file.replace('.h5','_new.h5')
ds = IncrementingHDF5Dataset(dest_file)
step = 100
dt = h5py.special_dtype(
vlen=str) # PY3 hdf5 datatype for variable-length Unicode strings
for i in range(0, len(L), step): #for i in range(0, 1000, 2000):
print('Processing: i=[' + str(i) + ':' + str(i + step) + ']')
these_indices = list(range(i, min(i + step, len(L))))
these_smiles = L[i:min(i + step, len(L))]
if grammar == 'new': # have to weed out non-parseable strings
tokens = [
my_model._tokenize(s.replace('-c', 'c')) for s in these_smiles
]
these_smiles, these_indices = list(
zip(*[(s, ind)
def train_policy_gradient(molecules=True,
grammar=True,
EPOCHS=None,
BATCH_SIZE=None,
reward_fun_on=None,
reward_fun_off=None,
max_steps=277,
lr_on=2e-4,
lr_discrim=1e-4,
p_thresh=0.5,
drop_rate=0.0,
plot_ignore_initial=0,
randomize_reward=False,
save_file=None,
reward_sm=0.0,
preload_file=None,
anchor_file=None,
anchor_weight=0.0,
decoder_type='action',
plot_prefix='',
dashboard='policy gradient',
smiles_save_file=None,
on_policy_loss_type='best',
off_policy_loss_type='mean',
sanity_checks=True):
root_location = os.path.dirname(
os.path.abspath(inspect.getfile(inspect.currentframe())))
root_location = root_location + '/../../'
gen_save_path = root_location + 'pretrained/gen_' + save_file
disc_save_path = root_location + 'pretrained/disc_' + save_file
if smiles_save_file is not None:
smiles_save_path = root_location + 'pretrained/' + smiles_save_file
save_dataset = IncrementingHDF5Dataset(smiles_save_path)
else:
save_dataset = None
settings = get_settings(molecules=molecules, grammar=grammar)
codec = get_codec(molecules, grammar, settings['max_seq_length'])
discrim_model = GraphDiscriminator(codec.grammar, drop_rate=drop_rate)
zinc_data = get_zinc_smiles()
zinc_set = set(zinc_data)
lookbacks = [BATCH_SIZE, 10 * BATCH_SIZE, 100 * BATCH_SIZE]
history_data = [deque(['O'], maxlen=lb) for lb in lookbacks]
def originality_mult(smiles_list):
out = []
for s in smiles_list:
if s in zinc_set:
out.append(0.5)
elif s in history_data[0]:
out.append(0.5)
elif s in history_data[1]:
out.append(0.70)
elif s in history_data[2]:
out.append(0.85)
else:
out.append(1.0)
return np.array(out)
def sigmoid(x):
tmp = -x #(
return 1 / (1 + np.exp(-x))
def discriminator_reward_mult(smiles_list):
orig_state = discrim_model.training
discrim_model.eval()
discrim_out_logits = discrim_model(smiles_list)['p_zinc']
discrim_probs = F.softmax(discrim_out_logits, dim=1)
prob_zinc = discrim_probs[:, 1].detach().cpu().numpy()
if orig_state:
discrim_model.train()
return prob_zinc
def adj_reward(x):
p = discriminator_reward_mult(x)
reward = np.maximum(reward_fun_on(x), 0)
out = reward * originality_mult(x) + 2 * p
return out
def adj_reward_old(x):
p = discriminator_reward_mult(x)
w = sigmoid(-(p - p_thresh) / 0.01)
if randomize_reward:
rand = np.random.uniform(size=p.shape)
w *= rand
reward = np.maximum(reward_fun_on(x), p_thresh)
weighted_reward = w * p + (1 - w) * reward
out = weighted_reward * originality_mult(x) #
return out
if EPOCHS is not None:
settings['EPOCHS'] = EPOCHS
if BATCH_SIZE is not None:
settings['BATCH_SIZE'] = BATCH_SIZE
task = SequenceGenerationTask(molecules=molecules,
grammar=grammar,
reward_fun=adj_reward,
batch_size=BATCH_SIZE,
max_steps=max_steps,
save_dataset=save_dataset)
model = get_decoder(molecules,
grammar,
z_size=settings['z_size'],
decoder_hidden_n=200,
feature_len=codec.feature_len(),
max_seq_length=max_steps,
drop_rate=drop_rate,
decoder_type=decoder_type,
task=task)[0]
# TODO: really ugly, refactor! In fact this model doesn't need a MaskingHead at all!
model.stepper.model.mask_gen.priors = True #'conditional' # use empirical priors for the mask gen
# if preload_file is not None:
# try:
# preload_path = root_location + 'pretrained/' + preload_file
# model.load_state_dict(torch.load(preload_path))
# except:
# pass
anchor_model = None
from generative_playground.molecules.rdkit_utils.rdkit_utils import NormalizedScorer
import rdkit.Chem.rdMolDescriptors as desc
import numpy as np
scorer = NormalizedScorer()
def model_process_fun(model_out, visdom, n):
# TODO: rephrase this to return a dict, instead of calling visdom directly
from rdkit import Chem
from rdkit.Chem.Draw import MolToFile
# actions, logits, rewards, terminals, info = model_out
smiles, valid = model_out['info']
total_rewards = model_out['rewards'].sum(1)
best_ind = torch.argmax(total_rewards).data.item()
this_smile = smiles[best_ind]
mol = Chem.MolFromSmiles(this_smile)
pic_save_path = root_location + 'images/' + 'tmp.svg'
if mol is not None:
try:
MolToFile(mol, pic_save_path, imageType='svg')
with open(pic_save_path, 'r') as myfile:
data = myfile.read()
data = data.replace('svg:', '')
visdom.append('best molecule of batch', 'svg', svgstr=data)
except Exception as e:
print(e)
scores, norm_scores = scorer.get_scores([this_smile])
visdom.append(
'score component',
'line',
X=np.array([n]),
Y=np.array(
[[x for x in norm_scores[0]] + [norm_scores[0].sum()] +
[scores[0].sum()] + [desc.CalcNumAromaticRings(mol)] +
[total_rewards[best_ind].item()]]),
opts={
'legend': [
'logP', 'SA', 'cycle', 'norm_reward', 'reward',
'Aromatic rings', 'eff_reward'
]
})
visdom.append('fraction valid',
'line',
X=np.array([n]),
Y=np.array([valid.mean().data.item()]))
if reward_fun_off is None:
reward_fun_off = reward_fun_on
# construct the loader to feed the discriminator
def make_callback(data):
def hc(inputs, model, outputs, loss_fn, loss):
graphs = outputs['graphs']
smiles = [g.to_smiles() for g in graphs]
for s in smiles: # only store unique instances of molecules so discriminator can't guess on frequency
if s not in data:
data.append(s)
return hc
# need to have something there to begin with, else the DataLoader constructor barfs
def get_rl_fitter(model,
loss_obj,
train_gen,
save_path,
fit_plot_prefix='',
model_process_fun=None,
lr=None,
extra_callbacks=[],
loss_display_cap=float('inf'),
anchor_model=None,
anchor_weight=0):
nice_params = filter(lambda p: p.requires_grad, model.parameters())
optimizer = optim.Adam(nice_params, lr=lr)
scheduler = lr_scheduler.StepLR(optimizer, step_size=100, gamma=0.99)
if dashboard is not None:
metric_monitor = MetricPlotter(
plot_prefix=fit_plot_prefix,
loss_display_cap=loss_display_cap,
dashboard_name=dashboard,
plot_ignore_initial=plot_ignore_initial,
process_model_fun=model_process_fun,
smooth_weight=0.9)
else:
metric_monitor = None
checkpointer = Checkpointer(valid_batches_to_checkpoint=1,
save_path=save_path,
save_always=True)
fitter = fit_rl(train_gen=train_gen,
model=model,
optimizer=optimizer,
scheduler=scheduler,
epochs=EPOCHS,
loss_fn=loss_obj,
grad_clip=5,
anchor_model=anchor_model,
anchor_weight=anchor_weight,
callbacks=[metric_monitor, checkpointer] +
extra_callbacks)
return fitter
class GeneratorToIterable:
def __init__(self, gen):
self.gen = gen
# we assume the generator is finite
self.len = 0
for _ in gen():
self.len += 1
def __len__(self):
return self.len
def __iter__(self):
return self.gen()
def my_gen():
for _ in range(1000):
yield to_gpu(torch.zeros(BATCH_SIZE, settings['z_size']))
# the on-policy fitter
history_callbacks = [make_callback(d) for d in history_data]
fitter1 = get_rl_fitter(model,
PolicyGradientLoss(on_policy_loss_type,
last_reward_wgt=reward_sm),
GeneratorToIterable(my_gen),
gen_save_path,
plot_prefix + 'on-policy',
model_process_fun=model_process_fun,
lr=lr_on,
extra_callbacks=history_callbacks,
anchor_model=anchor_model,
anchor_weight=anchor_weight)
#
# # get existing molecule data to add training
pre_dataset = EvenlyBlendedDataset(
2 * [history_data[0]] + history_data[1:],
labels=False) # a blend of 3 time horizons
dataset = EvenlyBlendedDataset([pre_dataset, zinc_data], labels=True)
discrim_loader = DataLoader(dataset, shuffle=True, batch_size=50)
celoss = nn.CrossEntropyLoss()
def my_loss(x):
# tmp = discriminator_reward_mult(x['smiles'])
# tmp2 = F.softmax(x['p_zinc'], dim=1)[:,1].detach().cpu().numpy()
# import numpy as np
# assert np.max(np.abs(tmp-tmp2)) < 1e-6
return celoss(x['p_zinc'].to(device), x['dataset_index'].to(device))
fitter2 = get_rl_fitter(discrim_model,
my_loss,
IterableTransform(
discrim_loader, lambda x: {
'smiles': x['X'],
'dataset_index': x['dataset_index']
}),
disc_save_path,
plot_prefix + ' discriminator',
lr=lr_discrim,
model_process_fun=None)
def on_policy_gen(fitter, model):
while True:
model.policy = SoftmaxRandomSamplePolicy(
) #bias=codec.grammar.get_log_frequencies())
yield next(fitter)
return model, on_policy_gen(fitter1, model), fitter2