import os import time import pickle import random from tqdm import tqdm import argparse import torch from torch_geometric.loader import DataLoader import torch.utils.data from rdkit import RDLogger torch.set_num_threads(5) RDLogger.DisableLog('rdApp.*') from utils import * from models import Generator from new_dataloader import DruggenDataset from loss import generator_loss from training_data import load_molecules class Inference(object): """Inference class for DrugGEN.""" def __init__(self, config): if config.set_seed: np.random.seed(config.seed) random.seed(config.seed) torch.manual_seed(config.seed) torch.cuda.manual_seed_all(config.seed) torch.backends.cudnn.deterministic = True torch.backends.cudnn.benchmark = False os.environ["PYTHONHASHSEED"] = str(config.seed) print(f'Using seed {config.seed}') self.device = torch.device("cuda" if torch.cuda.is_available() else 'cpu') # Initialize configurations self.submodel = config.submodel self.inference_model = config.inference_model self.sample_num = config.sample_num # Data loader. self.inf_raw_file = config.inf_raw_file # SMILES containing text file for first dataset. # Write the full path to file. self.inf_dataset_file = config.inf_dataset_file # Dataset file name for the first GAN. # Contains large number of molecules. self.inf_batch_size = config.inf_batch_size self.mol_data_dir = config.mol_data_dir # Directory where the dataset files are stored. self.dataset_name = self.inf_dataset_file.split(".")[0] self.max_atom = config.max_atom # Model is based on one-shot generation. # Max atom number for molecules must be specified. self.features = config.features # Small model uses atom types as node features. (Boolean, False uses atom types only.) # Additional node features can be added. Please check new_dataloarder.py Line 102. self.inf_dataset = DruggenDataset(self.mol_data_dir, self.inf_dataset_file, self.inf_raw_file, self.max_atom, self.features) # Dataset for the first GAN. Custom dataset class from PyG parent class. # Can create any molecular graph dataset given smiles string. # Nonisomeric SMILES are suggested but not necessary. # Uses sparse matrix representation for graphs, # For computational and speed efficiency. self.inf_loader = DataLoader(self.inf_dataset, shuffle=True, batch_size=self.inf_batch_size, drop_last=True) # PyG dataloader for the first GAN. # Atom and bond type dimensions for the construction of the model. self.atom_decoders = self.decoder_load("atom") # Atom type decoders for first GAN. # eg. 0:0, 1:6 (C), 2:7 (N), 3:8 (O), 4:9 (F) self.bond_decoders = self.decoder_load("bond") # Bond type decoders for first GAN. # eg. 0: (no-bond), 1: (single), 2: (double), 3: (triple), 4: (aromatic) self.m_dim = len(self.atom_decoders) if not self.features else int(self.inf_loader.dataset[0].x.shape[1]) # Atom type dimension. self.b_dim = len(self.bond_decoders) # Bond type dimension. self.vertexes = int(self.inf_loader.dataset[0].x.shape[0]) # Number of nodes in the graph. # Transformer and Convolution configurations. self.act = config.act self.dim = config.dim self.depth = config.depth self.heads = config.heads self.mlp_ratio = config.mlp_ratio self.dropout = config.dropout self.build_model() def build_model(self): """Create generators and discriminators.""" self.G = Generator(self.act, self.vertexes, self.b_dim, self.m_dim, self.dropout, dim=self.dim, depth=self.depth, heads=self.heads, mlp_ratio=self.mlp_ratio, submodel = self.submodel) self.print_network(self.G, 'G') self.G.to(self.device) def decoder_load(self, dictionary_name): ''' Loading the atom and bond decoders''' with open("data/decoders/" + dictionary_name + "_" + self.dataset_name + '.pkl', 'rb') as f: return pickle.load(f) def print_network(self, model, name): """Print out the network information.""" num_params = 0 for p in model.parameters(): num_params += p.numel() print(model) print(name) print("The number of parameters: {}".format(num_params)) def restore_model(self, submodel, model_directory): """Restore the trained generator and discriminator.""" print('Loading the model...') G_path = os.path.join(model_directory, '{}-G.ckpt'.format(submodel)) self.G.load_state_dict(torch.load(G_path, map_location=lambda storage, loc: storage)) def inference(self): # Load the trained generator. self.restore_model(self.submodel, self.inference_model) # smiles data for metrics calculation. chembl_smiles = [line for line in open("data/chembl_train.smi", 'r').read().splitlines()] chembl_test = [line for line in open("data/chembl_test.smi", 'r').read().splitlines()] drug_smiles = [line for line in open("data/akt_inhibitors.smi", 'r').read().splitlines()] drug_mols = [Chem.MolFromSmiles(smi) for smi in drug_smiles] drug_vecs = [AllChem.GetMorganFingerprintAsBitVect(x, 2, nBits=1024) for x in drug_mols if x is not None] # Make directories if not exist. if not os.path.exists("experiments/inference/{}".format(self.submodel)): os.makedirs("experiments/inference/{}".format(self.submodel)) self.G.eval() start_time = time.time() metric_calc_dr = [] uniqueness_calc = [] real_smiles_snn = [] nodes_sample = torch.Tensor(size=[1,45,1]).to(self.device) val_counter = 0 none_counter = 0 # Inference mode with torch.inference_mode(): pbar = tqdm(range(self.sample_num)) pbar.set_description('Inference mode for {} model started'.format(self.submodel)) for i, data in enumerate(self.inf_loader): val_counter += 1 # Preprocess dataset _, a_tensor, x_tensor = load_molecules( data=data, batch_size=self.inf_batch_size, device=self.device, b_dim=self.b_dim, m_dim=self.m_dim, ) _, _, node_sample, edge_sample = self.G(a_tensor, x_tensor) g_edges_hat_sample = torch.max(edge_sample, -1)[1] g_nodes_hat_sample = torch.max(node_sample, -1)[1] fake_mol_g = [self.inf_dataset.matrices2mol_drugs(n_.data.cpu().numpy(), e_.data.cpu().numpy(), strict=True, file_name=self.dataset_name) for e_, n_ in zip(g_edges_hat_sample, g_nodes_hat_sample)] a_tensor_sample = torch.max(a_tensor, -1)[1] x_tensor_sample = torch.max(x_tensor, -1)[1] real_mols = [self.inf_dataset.matrices2mol_drugs(n_.data.cpu().numpy(), e_.data.cpu().numpy(), strict=True, file_name=self.dataset_name) for e_, n_ in zip(a_tensor_sample, x_tensor_sample)] inference_drugs = [None if line is None else Chem.MolToSmiles(line) for line in fake_mol_g] inference_drugs = [None if x is None else max(x.split('.'), key=len) for x in inference_drugs] for molecules in inference_drugs: if molecules is None: none_counter += 1 with open("experiments/inference/{}/inference_drugs.txt".format(self.submodel), "a") as f: for molecules in inference_drugs: if molecules is not None: molecules = molecules.replace("*", "C") f.write(molecules) f.write("\n") uniqueness_calc.append(molecules) nodes_sample = torch.cat((nodes_sample, g_nodes_hat_sample.view(1,45,1)), 0) pbar.update(1) metric_calc_dr.append(molecules) generation_number = len([x for x in metric_calc_dr if x is not None]) if generation_number == self.sample_num or none_counter == self.sample_num: break real_smiles_snn.append(real_mols[0]) et = time.time() - start_time gen_vecs = [AllChem.GetMorganFingerprintAsBitVect(Chem.MolFromSmiles(x), 2, nBits=1024) for x in uniqueness_calc if Chem.MolFromSmiles(x) is not None] real_vecs = [AllChem.GetMorganFingerprintAsBitVect(x, 2, nBits=1024) for x in real_smiles_snn if x is not None] print("Inference mode is lasted for {:.2f} seconds".format(et)) print("Metrics calculation started using MOSES.") # post-process * to Carbon atom in valid molecules return{ "Runtime (seconds)": f"{et:.2f}", "Validity": f"{fraction_valid(metric_calc_dr):.2f}", "Uniqueness": f"{fraction_unique(uniqueness_calc):.2f}", "Novelty (Train)": f"{novelty(metric_calc_dr, chembl_smiles):.2f}", "Novelty (Inference)": f"{novelty(metric_calc_dr, chembl_test):.2f}", } if __name__=="__main__": parser = argparse.ArgumentParser() # Inference configuration. parser.add_argument('--submodel', type=str, default="DrugGEN", help="Chose model subtype: DrugGEN, NoTarget", choices=['DrugGEN', 'NoTarget']) parser.add_argument('--inference_model', type=str, help="Path to the model for inference") parser.add_argument('--sample_num', type=int, default=10000, help='inference samples') # Data configuration. parser.add_argument('--inf_dataset_file', type=str, default='chembl45_test.pt') parser.add_argument('--inf_raw_file', type=str, default='data/chembl_test.smi') parser.add_argument('--inf_batch_size', type=int, default=1, help='Batch size for inference') parser.add_argument('--mol_data_dir', type=str, default='data') parser.add_argument('--features', type=str2bool, default=False, help='features dimension for nodes') # Model configuration. parser.add_argument('--act', type=str, default="relu", help="Activation function for the model.", choices=['relu', 'tanh', 'leaky', 'sigmoid']) parser.add_argument('--max_atom', type=int, default=45, help='Max atom number for molecules must be specified.') parser.add_argument('--dim', type=int, default=128, help='Dimension of the Transformer Encoder model for the GAN.') parser.add_argument('--depth', type=int, default=1, help='Depth of the Transformer model from the GAN.') parser.add_argument('--heads', type=int, default=8, help='Number of heads for the MultiHeadAttention module from the GAN.') parser.add_argument('--mlp_ratio', type=int, default=3, help='MLP ratio for the Transformer.') parser.add_argument('--dropout', type=float, default=0., help='dropout rate') # Seed configuration. parser.add_argument('--set_seed', type=bool, default=False, help='set seed for reproducibility') parser.add_argument('--seed', type=int, default=1, help='seed for reproducibility') config = parser.parse_args() inference = Inference(config) inference.inference()