import numpy as np
from tqdm import tqdm
from typing import Optional, Union, Dict, Any, Tuple, List, Type
import torch
import torch.nn.functional as F
from torch_geometric.loader import DataLoader
from torch_geometric.data import Data
from .transformer import GraphTransformer
from .utils import PlaceHolder, to_dense, compute_dataset_info
from .diffusion import NoiseScheduleDiscrete, MarginalTransition, sample_discrete_features, sample_discrete_feature_noise, compute_batched_over0_posterior_distribution
from ...base import BaseMolecularGenerator
from ...utils import graph_from_smiles, graph_to_smiles
[docs]
class DigressMolecularGenerator(BaseMolecularGenerator):
"""
This generator implements the DiGress for unconditional molecular generation.
References
----------
- DiGress: Discrete Denoising diffusion for graph generation. International Conference on
Learning Representations (ICLR) 2023. https://openreview.net/forum?id=UaAD-Nu86WX
- Code: https://github.com/cvignac/DiGress
Parameters
----------
hidden_size_X : int, optional
Hidden dimension size for node features, defaults to 256
hidden_size_E : int, optional
Hidden dimension size for edge features, defaults to 128
num_layer : int, optional
Number of transformer layers, defaults to 5
n_head : int, optional
Number of attention heads, defaults to 8
dropout : float, optional
Dropout rate for transformer layers, defaults to 0.1
timesteps : int, optional
Number of diffusion timesteps, defaults to 500
batch_size : int, optional
Batch size for training, defaults to 128
epochs : int, optional
Number of training epochs, defaults to 10000
learning_rate : float, optional
Learning rate for optimization, defaults to 0.0002
grad_clip_value : Optional[float], optional
Value for gradient clipping (None = no clipping), defaults to None
weight_decay : float, optional
Weight decay for optimization, defaults to 0.0
lw_X : float, optional
Loss weight for node reconstruction, defaults to 1
lw_E : float, optional
Loss weight for edge reconstruction, defaults to 10
use_lr_scheduler : bool, optional
Whether to use learning rate scheduler, defaults to False
scheduler_factor : float, optional
Factor for learning rate scheduler (if use_lr_scheduler is True), defaults to 0.5
scheduler_patience : int, optional
Patience for learning rate scheduler (if use_lr_scheduler is True), defaults to 5
verbose : bool, optional
Whether to display progress bars and logs. Default is False.
device : Optional[Union[torch.device, str]], optional
Device to use for computation (cuda/cpu)
model_name : str, optional
Name of the model, defaults to "DigressMolecularGenerator"
"""
def __init__(
self,
hidden_size_X: int = 256,
hidden_size_E: int = 128,
num_layer: int = 5,
n_head: int = 8,
dropout: float = 0.1,
timesteps: int = 500,
batch_size: int = 512,
epochs: int = 1000,
learning_rate: float = 0.0002,
grad_clip_value: Optional[float] = None,
weight_decay: float = 1e-12,
lw_X: float = 1,
lw_E: float = 5,
use_lr_scheduler: bool = False,
scheduler_factor: float = 0.5,
scheduler_patience: int = 5,
verbose: bool = False,
device: Optional[Union[torch.device, str]] = None,
model_name: str = "DigressMolecularGenerator"
):
super().__init__(
device=device,
model_name=model_name,
)
self.hidden_size_X = hidden_size_X
self.hidden_size_E = hidden_size_E
self.num_layer = num_layer
self.n_head = n_head
self.dropout = dropout
self.timesteps = timesteps
self.batch_size = batch_size
self.epochs = epochs
self.learning_rate = learning_rate
self.grad_clip_value = grad_clip_value
self.weight_decay = weight_decay
self.lw_X = lw_X
self.lw_E = lw_E
self.use_lr_scheduler = use_lr_scheduler
self.scheduler_factor = scheduler_factor
self.scheduler_patience = scheduler_patience
self.verbose = verbose
self.fitting_loss = list()
self.fitting_epoch = 0
self.model_class = GraphTransformer
self.dataset_info = dict()
self.input_dim_X = None
self.input_dim_E = None
self.input_dim_y = 1
self.hidden_size_y = 128
self.max_node: Optional[int] = None
@staticmethod
def _get_param_names() -> List[str]:
return [
# Model Hyperparameters
"hidden_size_X", "hidden_size_E", "hidden_size_y", "num_layer", "n_head", "dropout",
"input_dim_X", "input_dim_E", "input_dim_y", "max_node",
# Diffusion parameters
"timesteps", "dataset_info",
# Training Parameters
"batch_size", "epochs", "learning_rate", "grad_clip_value",
"weight_decay", "lw_X", "lw_E",
# Scheduler Parameters
"use_lr_scheduler", "scheduler_factor", "scheduler_patience",
# Other Parameters
"fitting_epoch", "fitting_loss", "device", "verbose", "model_name"
]
def _get_model_params(self, checkpoint: Optional[Dict] = None) -> Dict[str, Any]:
params = ["num_layer", "input_dim_X", "input_dim_E", "input_dim_y",
"hidden_size_X", "hidden_size_E", "hidden_size_y", "n_head", "dropout"]
if checkpoint is not None:
if "hyperparameters" not in checkpoint:
raise ValueError("Checkpoint missing 'hyperparameters' key")
return {k: checkpoint["hyperparameters"][k] for k in params}
return {k: getattr(self, k) for k in params}
def _convert_to_pytorch_data(self, X, y=None):
"""Convert numpy arrays to PyTorch Geometric data format.
"""
if self.verbose:
iterator = tqdm(enumerate(X), desc="Converting molecules to graphs", total=len(X))
else:
iterator = enumerate(X)
pyg_graph_list = []
for idx, smiles_or_mol in iterator:
if y is not None:
properties = y[idx]
else:
properties = None
graph = graph_from_smiles(smiles_or_mol, properties)
g = Data()
# No H, first heavy atom has type 0
node_type = torch.from_numpy(graph['node_feat'][:, 0] - 1)
if node_type.numel() <= 1:
continue
# Filter out invalid node types (< 0)
valid_mask = node_type >= 0
if not valid_mask.all():
# Get valid nodes and adjust edge indices
valid_indices = torch.where(valid_mask)[0]
index_map = -torch.ones(node_type.size(0), dtype=torch.long)
index_map[valid_indices] = torch.arange(valid_indices.size(0))
# Filter edges that connect to invalid nodes
edge_index = torch.from_numpy(graph["edge_index"])
valid_edges_mask = valid_mask[edge_index[0]] & valid_mask[edge_index[1]]
valid_edge_index = edge_index[:, valid_edges_mask]
# Remap edge indices to account for removed nodes
remapped_edge_index = index_map[valid_edge_index]
# Filter edge attributes
edge_attr = torch.from_numpy(graph["edge_feat"])[:, 0] + 1
valid_edge_attr = edge_attr[valid_edges_mask]
# Update node and edge data
node_type = node_type[valid_mask]
g.edge_index = remapped_edge_index
g.edge_attr = valid_edge_attr.long()
else:
# No invalid nodes, proceed normally
g.edge_index = torch.from_numpy(graph["edge_index"])
edge_attr = torch.from_numpy(graph["edge_feat"])[:, 0] + 1
g.edge_attr = edge_attr.long()
# * is encoded as "misc" which is 119 - 1 and should be 117
node_type[node_type == 118] = 117
g.x = node_type.long()
# g.y = torch.from_numpy(graph["y"])
g.y = torch.zeros(1, 1)
del graph["node_feat"]
del graph["edge_index"]
del graph["edge_feat"]
del graph["y"]
pyg_graph_list.append(g)
return pyg_graph_list
def _setup_diffusion_params(self, X: Union[List, Dict]) -> None:
# Extract dataset info from X if it's a dict (from checkpoint), otherwise compute it
if isinstance(X, dict):
dataset_info = X["hyperparameters"]["dataset_info"]
timesteps = X["hyperparameters"]["timesteps"]
max_node = X["hyperparameters"]["max_node"]
else:
assert isinstance(X, list), "X must be a list of SMILES strings, but got {}".format(type(X))
dataset_info = compute_dataset_info(X)
timesteps = self.timesteps
max_node = dataset_info["max_node"]
self.input_dim_X = dataset_info["x_margins"].shape[0]
self.input_dim_E = dataset_info["e_margins"].shape[0]
self.dataset_info = dataset_info
self.timesteps = timesteps
self.max_node = max_node
x_limit = dataset_info["x_margins"].to(self.device)
e_limit = dataset_info["e_margins"].to(self.device)
self.transition_model = MarginalTransition(x_limit, e_limit, y_classes=self.input_dim_y)
self.limit_dist = PlaceHolder(X=x_limit, E=e_limit, y=None)
self.noise_schedule = NoiseScheduleDiscrete(timesteps=self.timesteps).to(self.device)
def _initialize_model(
self,
model_class: Type[torch.nn.Module],
checkpoint: Optional[Dict] = None
) -> torch.nn.Module:
"""Initialize the model with parameters or a checkpoint."""
model_params = self._get_model_params(checkpoint)
self.model = model_class(**model_params)
self.model = self.model.to(self.device)
if checkpoint is not None:
self._setup_diffusion_params(checkpoint)
self.model.load_state_dict(checkpoint["model_state_dict"])
return self.model
def _setup_optimizers(self) -> Tuple[torch.optim.Optimizer, Optional[Any]]:
"""Setup optimization components including optimizer and learning rate scheduler.
"""
optimizer = torch.optim.Adam(
self.model.parameters(), lr=self.learning_rate, weight_decay=self.weight_decay
)
scheduler = None
if self.use_lr_scheduler:
scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(
optimizer,
mode="min",
factor=self.scheduler_factor,
patience=self.scheduler_patience,
min_lr=1e-6,
cooldown=0,
eps=1e-8,
)
return optimizer, scheduler
[docs]
def fit(self, X_train: List[str]) -> "DigressMolecularGenerator":
num_task = 0 if self.input_dim_y is None else self.input_dim_y
X_train, _ = self._validate_inputs(X_train, num_task=num_task, return_rdkit_mol=False)
self._setup_diffusion_params(X_train)
self._initialize_model(self.model_class)
self.model.initialize_parameters()
optimizer, scheduler = self._setup_optimizers()
train_dataset = self._convert_to_pytorch_data(X_train)
train_loader = DataLoader(
train_dataset,
batch_size=self.batch_size,
shuffle=True,
num_workers=0
)
# Calculate total steps for global progress bar
steps_per_epoch = len(train_loader)
total_steps = self.epochs * steps_per_epoch
# Initialize global progress bar
global_pbar = None
if self.verbose:
global_pbar = tqdm(
total=total_steps,
desc="DiGress Training Progress",
unit="step",
dynamic_ncols=True,
leave=True
)
self.fitting_loss = []
self.fitting_epoch = 0
try:
for epoch in range(self.epochs):
train_losses = self._train_epoch(train_loader, optimizer, epoch, global_pbar)
epoch_loss = np.mean(train_losses).item()
self.fitting_loss.append(epoch_loss)
if scheduler:
scheduler.step(epoch_loss)
# Update global progress bar with epoch summary
if global_pbar is not None:
global_pbar.set_postfix({
"Epoch": f"{epoch+1}/{self.epochs}",
"Avg Loss": f"{epoch_loss:.4f}"
})
self.fitting_epoch = epoch
finally:
# Ensure progress bar is closed
if global_pbar is not None:
global_pbar.close()
self.is_fitted_ = True
return self
def _train_epoch(self, train_loader, optimizer, epoch, global_pbar=None):
"""Training logic for one epoch.
Args:
train_loader: DataLoader containing training data
optimizer: Optimizer instance for model parameter updates
epoch: Current epoch number
global_pbar: Global progress bar for tracking overall training progress
Returns:
list: List of loss values for each training step
"""
self.model.train()
losses = []
active_index = self.dataset_info["active_index"]
for step, batched_data in enumerate(train_loader):
batched_data = batched_data.to(self.device)
optimizer.zero_grad()
data_x = F.one_hot(batched_data.x, num_classes=118).float()[:, active_index]
data_edge_attr = F.one_hot(batched_data.edge_attr, num_classes=5).float()
dense_data, node_mask = to_dense(data_x, batched_data.edge_index, data_edge_attr, batched_data.batch, self.max_node)
dense_data = dense_data.mask(node_mask)
X, E = dense_data.X, dense_data.E
noisy_data = self.apply_noise(X, E, batched_data.y, node_mask)
loss, loss_X, loss_E = self.model.compute_loss(noisy_data, true_X=X, true_E=E, lw_X=self.lw_X, lw_E=self.lw_E)
if self.grad_clip_value is not None:
torch.nn.utils.clip_grad_norm_(self.model.parameters(), self.grad_clip_value)
loss.backward()
optimizer.step()
losses.append(loss.item())
# Update global progress bar
if global_pbar is not None:
global_pbar.update(1)
global_pbar.set_postfix({
"Epoch": f"{epoch+1}/{self.epochs}",
"Step": f"{step+1}/{len(train_loader)}",
"Loss": f"{loss.item():.4f}",
"Node Loss": f"{loss_X.item():.4f}",
"Edge Loss": f"{loss_E.item():.4f}"
})
return losses
def apply_noise(self, X, E, y, node_mask) -> Dict[str, Any]:
t_int = torch.randint(0, self.timesteps + 1, size=(X.size(0), 1), device=X.device).float() # (bs, 1)
s_int = t_int - 1
t_float = t_int / self.timesteps
s_float = s_int / self.timesteps
# beta_t and alpha_s_bar are used for denoising/loss computation
beta_t = self.noise_schedule(t_normalized=t_float) # (bs, 1)
alpha_s_bar = self.noise_schedule.get_alpha_bar(t_normalized=s_float) # (bs, 1)
alpha_t_bar = self.noise_schedule.get_alpha_bar(t_normalized=t_float) # (bs, 1)
Qtb = self.transition_model.get_Qt_bar(alpha_t_bar, device=self.device) # (bs, dx_in, dx_out), (bs, de_in, de_out)
assert (abs(Qtb.X.sum(dim=2) - 1.) < 1e-4).all(), Qtb.X.sum(dim=2) - 1
assert (abs(Qtb.E.sum(dim=2) - 1.) < 1e-4).all()
probX = X @ Qtb.X # (bs, n, dx_out)
probE = E @ Qtb.E.unsqueeze(1) # (bs, n, n, de_out)
sampled_t = sample_discrete_features(probX=probX, probE=probE, node_mask=node_mask)
X_t = F.one_hot(sampled_t.X, num_classes=self.input_dim_X)
E_t = F.one_hot(sampled_t.E, num_classes=self.input_dim_E)
assert (X.shape == X_t.shape) and (E.shape == E_t.shape)
z_t = PlaceHolder(X=X_t, E=E_t, y=y).type_as(X_t).mask(node_mask)
noisy_data = {'t': t_float * self.timesteps, 'beta_t': beta_t, 'alpha_s_bar': alpha_s_bar,
'alpha_t_bar': alpha_t_bar, 'X_t': z_t.X, 'E_t': z_t.E, 'y_t': z_t.y, 'node_mask': node_mask}
return noisy_data
[docs]
@torch.no_grad()
def generate(self, num_nodes: Optional[Union[List[List], np.ndarray, torch.Tensor]] = None, batch_size: int = 32) -> List[str]:
"""Randomly generate molecules with specified node counts.
Parameters
----------
num_nodes : Optional[Union[List[List], np.ndarray, torch.Tensor]], default=None
Number of nodes for each molecule in the batch. If None, samples from
the training distribution. Can be provided as:
- A list of lists
- A numpy array of shape (batch_size, 1)
- A torch tensor of shape (batch_size, 1)
batch_size : int, default=32
Number of molecules to generate.
Returns
-------
List[str]
List of generated molecules in SMILES format.
"""
if not self.is_fitted_:
raise ValueError("Model must be fitted before generating molecules.")
if self.input_dim_y is None or self.input_dim_X is None or self.input_dim_E is None or self.max_node is None:
raise ValueError(f"Model may not be fitted correctly as one of below attributes is not set: input_dim_y={self.input_dim_y}, input_dim_X={self.input_dim_X}, input_dim_E={self.input_dim_E}, max_node={self.max_node}")
if num_nodes is not None:
batch_size = len(num_nodes)
if num_nodes is None:
num_nodes_dist = self.dataset_info["num_nodes_dist"]
num_nodes = num_nodes_dist.sample_n(batch_size, self.device)
elif isinstance(num_nodes, list):
num_nodes = torch.tensor(num_nodes).to(self.device)
elif isinstance(num_nodes, np.ndarray):
num_nodes = torch.from_numpy(num_nodes).to(self.device)
if num_nodes.dim() == 1:
num_nodes = num_nodes.unsqueeze(-1)
assert num_nodes.size(0) == batch_size
arange = (
torch.arange(self.max_node).to(self.device)
.unsqueeze(0)
.expand(batch_size, -1)
)
node_mask = arange < num_nodes
if not hasattr(self, 'limit_dist') or self.limit_dist is None:
raise ValueError("Limit distribution not found. Please call setup_diffusion_params first.")
if not hasattr(self, 'dataset_info') or self.dataset_info is None:
raise ValueError("Dataset info not found. Please call setup_diffusion_params first.")
z_T = sample_discrete_feature_noise(
limit_dist=self.limit_dist, node_mask=node_mask
)
X, E = z_T.X, z_T.E
y = torch.zeros(batch_size, 1).to(self.device)
assert (E == torch.transpose(E, 1, 2)).all()
# Iteratively sample p(z_s | z_t) for t = 1, ..., T, with s = t - 1.
for s_int in reversed(range(0, self.timesteps)):
s_array = s_int * torch.ones((batch_size, 1)).float().to(self.device)
t_array = s_array + 1
s_norm = s_array / self.timesteps
t_norm = t_array / self.timesteps
# Sample z_s
sampled_s = self.sample_p_zs_given_zt(s_norm, t_norm, X, E, y, node_mask)
X, E, y = sampled_s.X, sampled_s.E, sampled_s.y
# Sample
sampled_s = sampled_s.mask(node_mask, collapse=True)
X, E, y = sampled_s.X, sampled_s.E, sampled_s.y
molecule_list = []
for i in range(batch_size):
n = num_nodes[i][0].item()
atom_types = X[i, :n].cpu()
edge_types = E[i, :n, :n].cpu()
molecule_list.append([atom_types, edge_types])
smiles_list = graph_to_smiles(molecule_list, self.dataset_info["atom_decoder"])
return smiles_list
def sample_p_zs_given_zt(
self, s, t, X_t, E_t, y, node_mask
):
"""Samples from zs ~ p(zs | zt). Only used during sampling.
if last_step, return the graph prediction as well"""
bs, n, _ = X_t.shape
beta_t = self.noise_schedule(t_normalized=t) # (bs, 1)
alpha_s_bar = self.noise_schedule.get_alpha_bar(t_normalized=s)
alpha_t_bar = self.noise_schedule.get_alpha_bar(t_normalized=t)
# Neural net predictions
noisy_data = {
"X_t": X_t,
"E_t": E_t,
"y_t": y,
"t": t,
"node_mask": node_mask,
}
pred = self.model(noisy_data)
# Normalize predictions
pred_X = F.softmax(pred.X, dim=-1) # bs, n, d0
pred_E = F.softmax(pred.E, dim=-1) # bs, n, n, d0
device = pred_X.device
# Retrieve transitions matrix
Qtb = self.transition_model.get_Qt_bar(alpha_t_bar, device)
Qsb = self.transition_model.get_Qt_bar(alpha_s_bar, device)
Qt = self.transition_model.get_Qt(beta_t, device)
p_s_and_t_given_0_X = compute_batched_over0_posterior_distribution(
X_t=X_t,
Qt=Qt.X,
Qsb=Qsb.X,
Qtb=Qtb.X
)
p_s_and_t_given_0_E = compute_batched_over0_posterior_distribution(
X_t=E_t,
Qt=Qt.E,
Qsb=Qsb.E,
Qtb=Qtb.E
)
# Dim of these two tensors: bs, N, d0, d_t-1
weighted_X = pred_X.unsqueeze(-1) * p_s_and_t_given_0_X # bs, n, d0, d_t-1
unnormalized_prob_X = weighted_X.sum(dim=2) # bs, n, d_t-1
unnormalized_prob_X[torch.sum(unnormalized_prob_X, dim=-1) == 0] = 1e-5
prob_X = unnormalized_prob_X / torch.sum(unnormalized_prob_X, dim=-1, keepdim=True) # bs, n, d_t-1
pred_E = pred_E.reshape((bs, -1, pred_E.shape[-1]))
weighted_E = pred_E.unsqueeze(-1) * p_s_and_t_given_0_E # bs, N, d0, d_t-1
unnormalized_prob_E = weighted_E.sum(dim=-2)
unnormalized_prob_E[torch.sum(unnormalized_prob_E, dim=-1) == 0] = 1e-5
prob_E = unnormalized_prob_E / torch.sum(unnormalized_prob_E, dim=-1, keepdim=True)
prob_E = prob_E.reshape(bs, n, n, pred_E.shape[-1])
assert ((prob_X.sum(dim=-1) - 1).abs() < 1e-4).all()
assert ((prob_E.sum(dim=-1) - 1).abs() < 1e-4).all()
sampled_s = sample_discrete_features(prob_X, prob_E, node_mask=node_mask)
X_s = F.one_hot(sampled_s.X, num_classes=self.input_dim_X).to(self.device).float()
E_s = F.one_hot(sampled_s.E, num_classes=self.input_dim_E).to(self.device).float()
assert (E_s == torch.transpose(E_s, 1, 2)).all()
assert (X_t.shape == X_s.shape) and (E_t.shape == E_s.shape)
out_one_hot = PlaceHolder(X=X_s, E=E_s, y=y)
return out_one_hot.mask(node_mask)