Targeting BCL-2 through Deep Learning-Based Drug Repurposing: A Multimodal Approach Combining Diffusion-Based Generative Modeling, Neural Relational Inference, and In Vitro Validation.
One block of code here has a permanent identifier (oscr:b:…), listed with the matches below: citable for good, with a page of its own. Select a match and press Cite to copy it. What a block identifier is.
The 1 match
- [1] § Methods › Chemprop Model Development for pIC50 Prediction ↔ wt/main.py, lines 1 to 86 · score 0.64 · random seed, trained model, hidden, batch, graph, predict · block
oscr:b:MD2Q5SPPASWR8-3· Report a mismatch
Paper
Loaded from Europe PMC by your browser; OSCR does not store it: doi.org · Europe PMC
The paper is loaded when this pane is shown.
The authors' code
Python · 482 lines · 20 KB · MIT · 1 match
- import time
- import argparse
- import pickle
- import os
- import datetime
- import torch
- import torch.optim as optim
- from torch.optim import lr_scheduler
- from utils import *
- from modules import *
- import copy
- import os
- parser = argparse.ArgumentParser(
- 'Neral relational inference for molecular dynamics simulations')
- parser.add_argument('--num-residues', type=int, default=352,
- help='Number of residues of the PDB.')
- parser.add_argument('--save-folder', type=str, default='logs',
- help='Where to save the trained model, leave empty to not save anything.')
- parser.add_argument('--load-folder', type=str, default='',
- help='Where to load the trained model if finetunning. ' +
- 'Leave empty to train from scratch')
- parser.add_argument('--edge-types', type=int, default=4,
- help='The number of edge types to infer.')
- parser.add_argument('--dims', type=int, default=6,
- help='The number of input dimensions used in study( position (X,Y,Z) + velocity (X,Y,Z) ). ')
- parser.add_argument('--timesteps', type=int, default=20,
- help='The number of time steps per sample. Actually is 50')
- parser.add_argument('--prediction-steps', type=int, default=1, metavar='N',
- help='Num steps to predict before re-using teacher forcing.')
- parser.add_argument('--no-cuda', action='store_true', default=False,
- help='Disables CUDA training.')
- parser.add_argument('--seed', type=int, default=42, help='Random seed.')
- parser.add_argument('--epochs', type=int, default=1000,
- help='Number of epochs to train.')
- parser.add_argument('--batch-size', type=int, default=1,
- help='Number of samples per batch.')
- parser.add_argument('--lr', type=float, default=0.0005,
- help='Initial learning rate.')
- parser.add_argument('--encoder-hidden', type=int, default=16,
- help='Number of hidden units in encoder.')
- parser.add_argument('--decoder-hidden', type=int, default=16,
- help='Number of hidden units in decoder.')
- parser.add_argument('--temp', type=float, default=0.5,
- help='Temperature for Gumbel softmax.')
- parser.add_argument('--encoder', type=str, default='mlp',
- help='Type of path encoder model (mlp or cnn).')
- parser.add_argument('--decoder', type=str, default='rnn',
- help='Type of decoder model (mlp, rnn, or sim).')
- parser.add_argument('--no-factor', action='store_true', default=False,
- help='Disables factor graph model.')
- parser.add_argument('--encoder-dropout', type=float, default=0.0,
- help='Dropout rate (1 - keep probability) in encoder.')
- parser.add_argument('--decoder-dropout', type=float, default=0.0,
- help='Dropout rate (1 - keep probability) in decoder.')
- parser.add_argument('--lr-decay', type=int, default=200,
- help='After how epochs to decay LR by a factor of gamma.')
- parser.add_argument('--gamma', type=float, default=0.5,
- help='LR decay factor.')
- parser.add_argument('--skip-first', action='store_true', default=True,
- help='Skip first edge type in decoder, i.e. it represents no-edge.')
- parser.add_argument('--var', type=float, default=5e-5,
- help='Output variance.')
- parser.add_argument('--hard', action='store_true', default=True,
- help='Uses discrete samples in training forward pass.')
- parser.add_argument('--prior', action='store_true', default=True,
- help='Whether to use sparsity prior.')
- parser.add_argument('--dynamic-graph', action='store_true', default=True,
- help='Whether test with dynamically re-computed graph.')
- parser.add_argument('--number-expstart', type=int, default=0,
- help='start number of experiments.')
- parser.add_argument('--number-exp', type=int, default=100,
- help='number of experiments.')
- args = parser.parse_args()
- args.cuda = not args.no_cuda and torch.cuda.is_available()
- args.factor = not args.no_factor
- # print all arguments
- print(args)
- np.random.seed(args.seed)
- torch.manual_seed(args.seed)
- if args.cuda:
- torch.cuda.manual_seed(args.seed)
- if args.dynamic_graph:
- print("Testing with dynamically re-computed graph.")
- # Save model and meta-data. Always saves in a new sub-folder.
- if args.save_folder:
- exp_counter = 0
- now = datetime.datetime.now()
- timestamp = now.isoformat()
- save_folder = args.save_folder+'/'
- if not os.path.isdir(save_folder):
- os.mkdir(save_folder)
- meta_file = os.path.join(save_folder, 'metadata.pkl')
- encoder_file = os.path.join(save_folder, 'encoder.pt')
- decoder_file = os.path.join(save_folder, 'decoder.pt')
- encoder_file_train = os.path.join(save_folder, 'encoder_train.pt')
- decoder_file_train = os.path.join(save_folder, 'decoder_train.pt')
- log_file = os.path.join(save_folder, 'log.txt')
- log = open(log_file, 'w')
- pickle.dump({'args': args}, open(meta_file, "wb"))
- log_file_train = os.path.join(save_folder, 'log_train.txt')
- log_train = open(log_file_train, 'w')
- else:
- print("WARNING: No save_folder provided!" +
- "Testing (within this script) will throw an error.")
- # load data
- train_loader, valid_loader, test_loader, loc_max, loc_min, vel_max, vel_min = load_dataset_train_valid_test(
- args.batch_size, args.number_exp, args.number_expstart, args.dims)
- # Generate off-diagonal interaction graph
- off_diag = np.ones([args.num_residues, args.num_residues]
- ) - np.eye(args.num_residues)
- rel_rec = np.array(encode_onehot(np.where(off_diag)[1]), dtype=np.float32)
- rel_send = np.array(encode_onehot(np.where(off_diag)[0]), dtype=np.float32)
- rel_rec = torch.FloatTensor(rel_rec)
- rel_send = torch.FloatTensor(rel_send)
- if args.encoder == 'mlp':
- encoder = MLPEncoder(args.timesteps * args.dims, args.encoder_hidden,
- args.edge_types,
- args.encoder_dropout, args.factor)
- elif args.encoder == 'cnn':
- encoder = CNNEncoder(args.dims, args.encoder_hidden,
- args.edge_types,
- args.encoder_dropout, args.factor)
- if args.decoder == 'mlp':
- decoder = MLPDecoder(n_in_node=args.dims,
- edge_types=args.edge_types,
- msg_hid=args.decoder_hidden,
- msg_out=args.decoder_hidden,
- n_hid=args.decoder_hidden,
- do_prob=args.decoder_dropout,
- skip_first=args.skip_first)
- elif args.decoder == 'rnn':
- decoder = RNNDecoder(n_in_node=args.dims,
- edge_types=args.edge_types,
- n_hid=args.decoder_hidden,
- do_prob=args.decoder_dropout,
- skip_first=args.skip_first)
- elif args.decoder == 'sim':
- decoder = SimulationDecoder(
- loc_max, loc_min, vel_max, vel_min, args.suffix)
- if args.load_folder:
- encoder_file = os.path.join('/media/arma/DATA/S-Ehsan/NRI_HITMER/wt/logs', 'encoder.pt')
- encoder.load_state_dict(torch.load(encoder_file))
- decoder_file = os.path.join(args.load_folder, 'decoder.pt')
- decoder.load_state_dict(torch.load(decoder_file))
- args.save_folder = False
- """
- encoder_file ='logs/encoder_train.pt'
- encoder.load_state_dict(torch.load(encoder_file))
- decoder_file = 'logs/decoder_train.pt'
- decoder.load_state_dict(torch.load(decoder_file))
- """
- optimizer = optim.Adam(list(encoder.parameters()) + list(decoder.parameters()),
- lr=args.lr)
- scheduler = lr_scheduler.StepLR(optimizer, step_size=args.lr_decay,
- gamma=args.gamma)
- # Linear indices of an upper triangular mx, used for acc calculation
- triu_indices = get_triu_offdiag_indices(args.num_residues)
- tril_indices = get_tril_offdiag_indices(args.num_residues)
- if args.prior:
- prior = np.array([0.91, 0.03, 0.03, 0.03]) # TODO: hard coded for now
- print("Using prior")
- print(prior)
- log_prior = torch.FloatTensor(np.log(prior))
- log_prior = torch.unsqueeze(log_prior, 0)
- log_prior = torch.unsqueeze(log_prior, 0)
- log_prior = Variable(log_prior)
- if args.cuda:
- log_prior = log_prior.cuda()
- if args.cuda:
- encoder.cuda()
- decoder.cuda()
- rel_rec = rel_rec.cuda()
- rel_send = rel_send.cuda()
- triu_indices = triu_indices.cuda()
- tril_indices = tril_indices.cuda()
- rel_rec = Variable(rel_rec)
- rel_send = Variable(rel_send)
- from time import process_time
- def train(epoch, best_val_loss,epochc):
- t = time.time()
- nll_train = []
- acc_train = []
- kl_train = []
- mse_train = []
- edges_train = []
- probs_train = []
- encoder.train()
- decoder.train()
- for batch_idx, (data, relations) in enumerate(train_loader):
- print(batch_idx)
- t1_start = process_time()
- if args.cuda:
- data, relations = data.cuda(), relations.cuda()
- data, relations = Variable(data), Variable(relations)
- optimizer.zero_grad()
- logits = encoder(data, rel_rec, rel_send)
- edges = gumbel_softmax(logits, tau=args.temp, hard=args.hard)
- prob = my_softmax(logits, -1)
- if args.decoder == 'rnn':
- output = decoder(data, edges, rel_rec, rel_send, 20,
- burn_in=True,
- burn_in_steps=args.timesteps - args.prediction_steps)
- else:
- output = decoder(data, edges, rel_rec, rel_send,
- args.prediction_steps)
- target = data[:, :, 1:, :]
- loss_nll = nll_gaussian(output, target, args.var)
- if args.prior:
- loss_kl = kl_categorical(prob, log_prior, args.num_residues)
- else:
- loss_kl = kl_categorical_uniform(prob, args.num_residues,
- args.edge_types)
- loss = loss_nll + loss_kl
- acc = edge_accuracy(logits, relations)
- acc_train.append(acc)
- loss.backward()
- optimizer.step()
- mse_train.append(F.mse_loss(output, target).item())
- nll_train.append(loss_nll.item())
- kl_train.append(loss_kl.item())
- _, edges_t = edges.max(-1)
- edges_train.append(edges_t.data.cpu().numpy())
- probs_train.append(prob.data.cpu().numpy())
- t1_stop = process_time()
- print("Elapsed time during the whole program in seconds:",
- t1_stop-t1_start)
- scheduler.step()
- nll_val = []
- acc_val = []
- kl_val = []
- mse_val = []
- encoder.eval()
- decoder.eval()
- if epochc%10==0:
- for batch_idx, (data, relations) in enumerate(valid_loader):
- if args.cuda:
- data, relations = data.cuda(), relations.cuda()
- with torch.no_grad():
- logits = encoder(data, rel_rec, rel_send)
- edges = gumbel_softmax(logits, tau=args.temp, hard=True)
- prob = my_softmax(logits, -1)
- # validation output uses teacher forcing
- output = decoder(data, edges, rel_rec, rel_send, 1)
- target = data[:, :, 1:, :]
- loss_nll = nll_gaussian(output, target, args.var)
- loss_kl = kl_categorical_uniform(
- prob, args.num_residues, args.edge_types)
- acc = edge_accuracy(logits, relations)
- acc_val.append(acc)
- mse_val.append(F.mse_loss(output, target).item())
- nll_val.append(loss_nll.item())
- kl_val.append(loss_kl.item())
- if args.save_folder and np.mean(np.array(nll_val)) < best_val_loss:
- torch.save(encoder.state_dict(), encoder_file)
- torch.save(decoder.state_dict(), decoder_file)
- print('Best model so far, saving...')
- print('Epoch: {:04d}'.format(epoch),
- 'nll_train: {:.10f}'.format(np.mean(np.array(nll_train))),
- 'kl_train: {:.10f}'.format(np.mean(np.array(kl_train))),
- 'mse_train: {:.10f}'.format(np.mean(np.array(mse_train))),
- 'acc_train: {:.10f}'.format(np.mean(np.array(acc_train))),
- 'nll_val: {:.10f}'.format(np.mean(np.array(nll_val))),
- 'kl_val: {:.10f}'.format(np.mean(np.array(kl_val))),
- 'mse_val: {:.10f}'.format(np.mean(np.array(mse_val))),
- 'acc_val: {:.10f}'.format(np.mean(np.array(acc_val))),
- 'time: {:.4f}s'.format(time.time() - t), file=log)
- log.flush()
- print('Epoch: {:04d}'.format(epoch),
- 'nll_train: {:.10f}'.format(np.mean(np.array(nll_train))),
- 'kl_train: {:.10f}'.format(np.mean(np.array(kl_train))),
- 'mse_train: {:.10f}'.format(np.mean(np.array(mse_train))),
- 'acc_train: {:.10f}'.format(np.mean(np.array(acc_train))),
- 'nll_val: {:.10f}'.format(np.mean(np.array(nll_val))),
- 'kl_val: {:.10f}'.format(np.mean(np.array(kl_val))),
- 'mse_val: {:.10f}'.format(np.mean(np.array(mse_val))),
- 'acc_val: {:.10f}'.format(np.mean(np.array(acc_val))),
- 'time: {:.4f}s'.format(time.time() - t))
- edges_train = np.concatenate(edges_train)
- probs_train = np.concatenate(probs_train)
- print('Epoch: {:04d}'.format(epoch),
- 'nll_train: {:.10f}'.format(np.mean(np.array(nll_train))),
- 'kl_train: {:.10f}'.format(np.mean(np.array(kl_train))),
- 'mse_train: {:.10f}'.format(np.mean(np.array(mse_train))),
- 'acc_train: {:.10f}'.format(np.mean(np.array(acc_train))),
- 'nll_val: {:.10f}'.format(np.mean(np.array(nll_val))),
- 'kl_val: {:.10f}'.format(np.mean(np.array(kl_val))),
- 'mse_val: {:.10f}'.format(np.mean(np.array(mse_val))),
- 'acc_val: {:.10f}'.format(np.mean(np.array(acc_val))),
- 'time: {:.4f}s'.format(time.time() - t), file=log_train)
- log_train.flush()
- torch.save(encoder.state_dict(), encoder_file_train)
- torch.save(decoder.state_dict(), decoder_file_train)
- return encoder, decoder, edges_train, probs_train, np.mean(np.array(nll_val))
- def test():
- acc_test = []
- nll_test = []
- kl_test = []
- mse_test = []
- edges_test = []
- probs_test = []
- tot_mse = 0
- counter = 0
- encoder.eval()
- decoder.eval()
- encoder.load_state_dict(torch.load(encoder_file))
- decoder.load_state_dict(torch.load(decoder_file))
- for batch_idx, (data, relations) in enumerate(test_loader):
- if args.cuda:
- data, relations = data.cuda(), relations.cuda()
- with torch.no_grad():
- # assert (data.size(2) - args.timesteps) >= args.timesteps
- assert (data.size(2)) >= args.timesteps
- data_encoder = data[:, :, :args.timesteps, :].contiguous()
- data_decoder = data[:, :, -args.timesteps:, :].contiguous()
- logits = encoder(data_encoder, rel_rec, rel_send)
- edges = gumbel_softmax(logits, tau=args.temp, hard=True)
- prob = my_softmax(logits, -1)
- output = decoder(data_decoder, edges, rel_rec, rel_send, 1)
- target = data_decoder[:, :, 1:, :]
- loss_nll = nll_gaussian(output, target, args.var)
- loss_kl = kl_categorical_uniform(
- prob, args.num_residues, args.edge_types)
- acc = edge_accuracy(logits, relations)
- acc_test.append(acc)
- mse_test.append(F.mse_loss(output, target).item())
- nll_test.append(loss_nll.item())
- kl_test.append(loss_kl.item())
- _, edges_t = edges.max(-1)
- edges_test.append(edges_t.data.cpu().numpy())
- probs_test.append(prob.data.cpu().numpy())
- # For plotting purposes
- if args.decoder == 'rnn':
- if args.dynamic_graph:
- output = decoder(data, edges, rel_rec, rel_send, 20,
- burn_in=False, burn_in_steps=args.timesteps,
- dynamic_graph=True, encoder=encoder,
- temp=args.temp)
- else:
- output = decoder(data, edges, rel_rec, rel_send, 20,
- burn_in=True, burn_in_steps=args.timesteps)
- target = data[:, :, 1:, :]
- else:
- data_plot = data[:, :, 0:0 + 21,
- :].contiguous()
- output = decoder(data_plot, edges, rel_rec, rel_send, 20)
- target = data_plot[:, :, 1:, :]
- mse = ((target - output) ** 2).mean(dim=0).mean(dim=0).mean(dim=-1)
- tot_mse += mse.data.cpu().numpy()
- counter += 1
- mean_mse = tot_mse / counter
- mse_str = '['
- for mse_step in mean_mse[:-1]:
- mse_str += " {:.12f} ,".format(mse_step)
- mse_str += " {:.12f} ".format(mean_mse[-1])
- mse_str += ']'
- print('--------------------------------')
- print('--------Testing-----------------')
- print('--------------------------------')
- print('nll_test: {:.10f}'.format(np.mean(nll_test)),
- 'kl_test: {:.10f}'.format(np.mean(kl_test)),
- 'mse_test: {:.10f}'.format(np.mean(mse_test)),
- 'acc_test: {:.10f}'.format(np.mean(acc_test)))
- print('MSE: {}'.format(mse_str))
- edges_test = np.concatenate(edges_test)
- probs_test = np.concatenate(probs_test)
- if args.save_folder:
- print('--------------------------------', file=log)
- print('--------Testing-----------------', file=log)
- print('--------------------------------', file=log)
- print('nll_test: {:.10f}'.format(np.mean(nll_test)),
- 'kl_test: {:.10f}'.format(np.mean(kl_test)),
- 'mse_test: {:.10f}'.format(np.mean(mse_test)),
- 'acc_test: {:.10f}'.format(np.mean(acc_test)),
- file=log)
- print('MSE: {}'.format(mse_str), file=log)
- log.flush()
- return edges_test, probs_test
- # Train model
- print("Start Training...")
- t_total = time.time()
- best_val_loss = np.inf
- best_epoch = 0
- for epoch in range(args.epochs):
- """
- if epoch==0:
- best_enc_wts = copy.deepcopy(encoder.state_dict())
- best_dec_wts = copy.deepcopy(decoder.state_dict())
- """
- encoder, decoder, edges_train, probs_train, val_loss = train(
- epoch, best_val_loss,epoch)
- # print('Epoch '+str(epoch)+' with val loss:'+str(val_loss))
- np.save(str(args.save_folder)+'/out_edges_train.npy', edges_train)
- np.save(str(args.save_folder)+'/out_probs_train.npy', probs_train)
- if val_loss < best_val_loss:
- best_val_loss = val_loss
- best_epoch = epoch
- best_enc_wts = copy.deepcopy(encoder.state_dict())
- best_dec_wts = copy.deepcopy(decoder.state_dict())
- np.save(str(args.save_folder)+'/out_edges_val.npy', edges_train)
- np.save(str(args.save_folder)+'/out_probs_val.npy', probs_train)
- """
- encoder.load_state_dict(best_enc_wts)
- decoder.load_state_dict(best_dec_wts)
- """
- np.save(str(args.save_folder)+'/out_edges_train.npy', edges_train)
- np.save(str(args.save_folder)+'/out_probs_train.npy', probs_train)
- print("Optimization Finished!")
- print("Best Epoch: {:04d}".format(best_epoch))
- if args.save_folder:
- print("Best Epoch: {:04d}".format(best_epoch), file=log)
- log.flush()
- # Test
- edges_test, probs_test = test()
- if log is not None:
- print(save_folder)
- log.close()
main.py at commit 6b4f7bd, under MIT · at the source
Overview
- Lab for Innovative Drugs (Lab4IND), Computational Drug Design Center (HİTMER), Bahçeşehir University, İstanbul 34349, Türkiye
- Department of Biostatistics and Medical Informatics, School of Medicine, Bahçeşehir University, Istanbul 34349, Türkiye
- Department of Medical Biology, School of Medicine, Bahçeşehir University, Istanbul 34349, Türkiye
- Molecular Therapy Lab, Department of Pharmaceutical Chemistry, School of Pharmacy, Bahçeşehir University, Istanbul 34353, Türkiye
- Quantitative System Biology Lab, Faculty of Medicine, Biruni University, İstanbul 34010, Türkiye
Abstract
Accurate identification of repurposable BCL-2 ligands requires not only plausible bound complex structures but also a dynamic description of how ligand binding reshapes residue-level communication. Here, we present a multimodal BCL-2 repurposing workflow built with diffusion-based generative modeling for ligand-specific complex generation and an extended neural relational inference (NRI) framework for trajectory-level interaction analysis. NeuralPlexer was applied to a library of 3094 FDA-approved drugs to generate BCL-2-ligand complex conformations at scale, yielding 1294 structurally acceptable complexes for downstream prioritization. To complement static scoring, filtered candidates were evaluated by molecular docking, anticancer QSAR classification, all-atom molecular dynamics (MD) simulations, and MM/
Reproduced under the paper's license (CC BY), from the paper cited above.
Repositories
Its files are read in the Code ↔ Paper reader above, with 1 match between paragraphs and lines of code.
DurdagiLab/Neuralplexer_Ligand_Scoring
7057748744fc580fa1fc1052a26bff7649769409, 18 July 2024Availability: 1 check, the latest on 26 September 2026: the link answers
- 26 September 2026: the link answers
4 files
- Neuralplexer_get_RANK1_p
rotein.py , Python, 97 lines - neuralplexer_rank1_score
_inplace.py , Python, 52 lines - LICENSE, License, 21 lines
- README.md, Text, 27 lines
DurdagiLab/Automate-Neural-relational-inference-NRI-
6b4f7bd16cba9fb8e963d11e2aafc2a42d96f597, 18 July 2024Availability: 1 check, the latest on 26 September 2026: the link answers
- 26 September 2026: the link answers
13 files
- Auto_NRI.py, Python, 209 lines
- wt/
Test_Trajectory.py , Python, 252 lines - wt/
__init__.py , Python, 1 line - wt/
convert_dataset.py , Python, 378 lines - wt/
main.py , Python, 482 lines, 1 match - wt/
modules.py , Python, 655 lines - wt/
postanalysis_path.py , Python, 142 lines - wt/
postanalysis_visual.py , Python, 208 lines - wt/
postanalysis_visual_val. , Python, 199 linespy - wt/
utils.py , Python, 579 lines - wt/
visual.py , Python, 36 lines - LICENSE, License, 21 lines
- README.md, Text, 36 lines
Tracing map
Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.
What the map holds:
- 2 repositories of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
- 13 scripts, each with its path and the digest of its content;
- 1 match between paragraphs of the paper and lines of the code (method lexical-v1);
- neither the text of the paper nor the code itself.
Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.
Data
No dataset and no data link were found in the paper.
Data Availability Statement
The simulations were conducted with Desmond program and the MD data collected as trajectory files, and NeuralPlexer structures of top candidates were made available via the following repository: https://
Reproduced under the paper's license (CC BY), from the paper cited above.
Versions
The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.
Version 3, 28 September 2026
- Publisher: n/a → American Chemical Society
Version 1, 27 September 2026: the first record
Recorded: type, language, journal, volume, issue, pages, dates, 5 authors, 11 MeSH terms, 2 funders, 32 references.
Cite
This paper
Sayyah, E., Tunç, H., Çelebi, A., Avşar, T., & Durdağı, S. (2026). Targeting BCL-2 through Deep Learning-Based Drug Repurposing: A Multimodal Approach Combining Diffusion-Based Generative Modeling, Neural Relational Inference, and In Vitro Validation. Journal of chemical information and modeling, 66(17), 11361-11376. https://
BibTeX
@article{sayyah2026targe
author = {Sayyah, Ehsan and Tunç, Hüseyin and Çelebi, Asuman and Avşar, Timuçin and Durdağı, Serdar},
title = {{Targeting BCL-2 through Deep Learning-Based Drug Repurposing: A Multimodal Approach Combining Diffusion-Based Generative Modeling, Neural Relational Inference, and In Vitro Validation}},
journal = {Journal of chemical information and modeling},
year = {2026},
month = sep,
volume = {66},
number = {17},
pages = {11361--11376},
publisher = {American Chemical Society},
issn = {1549-9596},
doi = {10.1021/
url = {https://
pmid = {42734502},
pmcid = {PMC13580121}
}
RIS
TY - JOUR
AU - Sayyah, Ehsan
AU - Tunç, Hüseyin
AU - Çelebi, Asuman
AU - Avşar, Timuçin
AU - Durdağı, Serdar
TI - Targeting BCL-2 through Deep Learning-Based Drug Repurposing: A Multimodal Approach Combining Diffusion-Based Generative Modeling, Neural Relational Inference, and In Vitro Validation
T2 - Journal of chemical information and modeling
J2 - J Chem Inf Model
PY - 2026
DA - 2026/
VL - 66
IS - 17
SP - 11361
EP - 11376
SN - 1549-9596
PB - American Chemical Society
DO - 10.1021/
UR - https://
LA - en
ER -
CSL-JSON
{
"id": "10.1021/
"type": "article-journal",
"title": "Targeting BCL-2 through Deep Learning-Based Drug Repurposing: A Multimodal Approach Combining Diffusion-Based Generative Modeling, Neural Relational Inference, and In Vitro Validation",
"container-title": "Journal of chemical information and modeling",
"author": [
{
"family": "Sayyah",
"given": "Ehsan"
},
{
"family": "Tunç",
"given": "Hüseyin"
},
{
"family": "Çelebi",
"given": "Asuman"
},
{
"family": "Avşar",
"given": "Timuçin"
},
{
"family": "Durdağı",
"given": "Serdar"
}
],
"container-title-short":
"volume": "66",
"issue": "17",
"page": "11361-11376",
"DOI": "10.1021/
"PMID": "42734502",
"PMCID": "PMC13580121",
"ISSN": "1549-9596",
"publisher": "American Chemical Society",
"URL": "https://
"language": "en",
"issued": {
"date-parts": [
[
2026,
9,
1
]
]
}
}
The tracing map gets a citation of its own once an author has validated it and it has a DOI.
Similar papers
The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.
- [1] doi:10.1021/acschemneuro.5c00519
- Niraparib Demonstrates Therapeutic Potential in Multiple Sclerosis through Inhibition of IL-17A Receptor Interaction and Promotion of Remyelination.Journal: ACS chemical neuroscienceIn common: cellular / molecular, 6 references, 2 authors
- [2] doi:10.1371/journal.pcbi.1013966 [code]
- Carba-NAD binding activates SIR2 by reshaping conformational plasticity and rewiring long-range allosteric networks.Journal: PLoS computational biologyIn common: RDKit, NetworkX, PyTorch, 5 other tools, computational modeling (no new data), 1 reference
- [3] doi:10.1038/s41467-023-41343-1 [code]
- Machine learning coarse-grained potentials of protein thermodynamics.Journal: Nature communicationsIn common: RDKit, NetworkX, PyTorch, 5 other tools, computational modeling (no new data), other condition, cellular / molecular, 1 reference
- [4] doi:10.1371/journal.pcbi.1014879 [code]
- Evolutionary and dynamic constraints converge on a conserved coupling sector in thermally adapted GH1 β-glucosidases.Journal: PLoS computational biologyIn common: NetworkX, PyTorch, seaborn, 4 other tools, computational modeling (no new data), 2 references
- [5] doi:10.1038/s41467-025-60963-3 [code]
- Zero shot molecular generation via similarity kernels.Journal: Nature communicationsIn common: RDKit, NetworkX, PyTorch, 5 other tools, cellular / molecular, 1 reference
- [6] doi:10.3389/fbinf.2024.1381540 [code]
- The evolution of mammalian Rem2: unraveling the impact of purifying selection and coevolution on protein function, and implications for human disorders.Journal: Frontiers in bioinformaticsIn common: RDKit, NetworkX, PyTorch, 5 other tools, other condition, cellular / molecular, 1 reference
- [7] doi:10.1038/s41467-025-64392-0 [code]
- Structure-guided discovery of Otopetrin 1 inhibitors reveals druggable binding sites at the intrasubunit interface.Journal: Nature communicationsIn common: RDKit, NetworkX, PyTorch, 5 other tools, cellular / molecular, 1 reference
- [8] doi:10.1038/s41467-024-55471-9 [code]
- Structural insights into the mechanism of phosphate recognition and transport by XPR1.Journal: Nature communicationsIn common: RDKit, NetworkX, PyTorch, 5 other tools, cellular / molecular, 1 reference
- [9] doi:10.1093/bib/bbag118 [code]
- Drug screening for α-synuclein aggregation inhibitors via multimodal graph neural network.Journal: Briefings in bioinformaticsIn common: RDKit, NetworkX, PyTorch, 5 other tools, computational modeling (no new data), cellular / molecular
- [10] doi:10.1093/narcan/zcaf029 [code]
- Neural interaction explainable AI predicts drug response across cancers.Journal: NAR cancerIn common: RDKit, NetworkX, PyTorch, 5 other tools, other condition, cellular / molecular
Contribute
The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.
Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.
Claim this paper
Correct its record
Say what each link of this record is, remove the ones that are not the paper's, add the ones that are missing. The correction becomes a new version of the record, in its Versions section.
Validate its tracing map
You validate the map as this page shows it: 2 repositories of the authors' code, each at its verified commit and with its license, 13 scripts, and 1 match between paragraphs and code (see the Code and Map sections). It then receives a DOI on Zenodo, with you (your ORCID iD) and OSCR as its creators; the code itself is not deposited.
The map's fingerprint: sha256:2552470c1ba32677…
Add the badge to the code's README
If you are an author or a maintainer of this paper's code, you can add this badge. It says what OSCR knows of the paper (its code, its validated tracing map) and links the code to this page. Paste one of these snippets under the README's title. Keep the two marks around it, so that the badge can be replaced in place and is never added twice. OSCR changes nothing in your repository.
Markdown
<!-- oscr-badge:start -->
[](https://openscicode.org/badge/doi/10.1021/acs.jcim.6c01299)
<!-- oscr-badge:end -->reStructuredText
.. oscr-badge:start
.. image:: https://openscicode.org/badge/doi/10.1021/acs.jcim.6c01299.svg
:target: https://openscicode.org/badge/doi/10.1021/acs.jcim.6c01299
:alt: OSCR
.. oscr-badge:endHTML
<!-- oscr-badge:start -->
<a href="https://openscicode.org/badge/doi/10.1021/acs.jcim.6c01299"><img src="https://openscicode.org/badge/doi/10.1021/acs.jcim.6c01299.svg" alt="OSCR"></a>
<!-- oscr-badge:end -->On GitHub, open the README, choose Edit (the pencil), paste the snippet under the title, then "Commit changes".
- DurdagiLab/Neuralplexer_Ligand_Scoring: README.md on GitHub
- DurdagiLab/Automate-Neural-relational-inference-NRI-: README.md on GitHub
If you choose to, you can have the badge kept up to date for you, with oscr badge --write from the command line or with the GitHub Action in your repository's workflows. Every other form and option (a card, AsciiDoc, LaTeX, shields.io) is on the badge page.
Request its removal
To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).
Discussion, reproductions, activity
Discussion: code errors and mismatches between the code and the paper about this paper and its code, from readers and its authors, beside the paper and its tracing map.
Report a mismatch between the code and the paper · Report a code error (signed in)
Reproductions: reports from readers who ran the authors' code and did not get the paper's result, what they ran, with which environment, commit and data.
Report a reproduction failure (signed in)
Versions of its code: the releases of its code tied to a version of this paper (the preprint, the accepted manuscript, the version of record…), each with the tracing map versioned with it.
Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.
