PVAED: prior-guided variational autoencoders with diffusion denoising for interpretable single-cell representation learning.
3 blocks of code here have 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 6 matches
- [1] § Materials and methods › Clustering metrics ↔ PVAED/down_stream.ipynb, lines 31 to 48 · score 0.84 · Adjusted Mutual, Adjusted Rand, Normalized Mutual, AMI, Homogeneity, ARI · Report a mismatch
- [2] § Materials and methods › PVAED loss function ↔ PVAED/main.py, lines 142 to 182 · score 0.64 · KL divergence, reconstruction loss, neighbors, trustworthiness · block
oscr:b:1377122Z2VP6K-T· Report a mismatch - [3] § Materials and methods › PVAED loss function ↔ PVAED/main_joint_fine_tunning.py, lines 161 to 185 · score 0.64 · KL divergence, reconstruction loss, neighbors, trustworthiness · block
oscr:b:BCRAJHBV8VRKF-F· Report a mismatch - [4] § Materials and methods › PVAED loss function ↔ PVAED/main.py, lines 142 to 182 · score 0.63 · reconstruction loss, loss function, Gaussian, KL, divergence, Trustworthiness · block
oscr:b:1377122Z2VP6K-T· Report a mismatch - [5] § Materials and methods › PVAED loss function ↔ PVAED/main_joint_fine_tunning.py, lines 161 to 185 · score 0.63 · reconstruction loss, loss function, Gaussian, KL, divergence, Trustworthiness · block
oscr:b:BCRAJHBV8VRKF-F· Report a mismatch - [6] § Materials and methods › Prior knowledge informed architectural design of PVAED ↔ PVAED/main_joint_fine_tunning.py, lines 38 to 70 · score 0.58 · joint fine tuning, gene interactions, DDPM, diffusion, encoder, VAE · block
oscr:b:B4PC85YXJ4WPY-0· 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 · 387 lines · 17 KB · no license · 3 matches
- import torch
- from torch import optim
- import torch.nn.functional as F
- from torch.utils.data import DataLoader
- from torchvision import transforms, datasets
- from torchvision.utils import save_image
- from vae import VAE
- import matplotlib.pyplot as plt
- import argparse
- import os
- import shutil
- import numpy as np
- import pandas as pd
- import openpyxl
- import scanpy as sc
- import time
- import networkx as nx
- import torch.nn as nn
- from scipy.sparse import csr_matrix
- from scipy.sparse import save_npz
- from scipy.sparse import load_npz
- from scipy.stats import spearmanr
- import DDPM
- # plt.style.use("ggplot")
- from DDPM import build_network
- from DDPM import train
- from DDPM import DDPM
- import sklearn
- from sklearn.manifold import trustworthiness
- from utils_joint import save_file
- cuda = torch.cuda.is_available()
- device = torch.device("cuda" if cuda else "cpu")
- #device = 'cuda'
- parser = argparse.ArgumentParser(description="Variational Auto-Encoder MNIST Example")
- parser.add_argument('--dataset_name', type=str, default='data_GSE204684_developing_human_cerebral_cortex', metavar='N', help='dataset name')
- args_ini = parser.parse_args()
- parser.add_argument('--vae_pre_dir', type=str, default='./%s/checkPoint'%args_ini.dataset_name, metavar='N', help='model saving directory')
- parser.add_argument('--vae_pre_result_dir', type=str, default='./%s/VAEResult'%args_ini.dataset_name, metavar='DIR', help='output directory')
- parser.add_argument('--vae_after_dir', type=str, default='./%s/vae_after/checkPoint'%args_ini.dataset_name, metavar='N', help='model saving directory')
- parser.add_argument('--vae_after_result_dir', type=str, default='./%s/vae_after/VAEResult'%args_ini.dataset_name, metavar='N', help='model saving directory')
- parser.add_argument('--scData_dir', type=str, default='./data_hvg5000.h5ad', metavar='DIR', help='scData directory')
- parser.add_argument('--fine_scData_dir', type=str, default='./data_hvg5000.h5ad', metavar='DIR', help='scData directory')
- parser.add_argument('--pathway_dir', type=str, default='./prior_data_results/pathway_use_allgenes_maskedmatrix_filtered_geneNumberOver5.csv', metavar='DIR', help='pathway directory')
- parser.add_argument('--complex_dir', type=str, default='./prior_data_results/complex_use_allgenes_maskedmatrix_filtered_geneNumberOver5.csv', metavar='DIR', help='complex directory')
- parser.add_argument('--TF_dir', type=str, default='./prior_data_results/TF_use_allgenes_maskedmatrix_filtered_geneNumberOver5.csv', metavar='DIR', help='TF directory')
- parser.add_argument('--adjMatrix_kegg_dir', type=str, default='./prior_data_results/pathway_common_with_KEGGgraph_adj_matrix_Sparse.npz', metavar='DIR', help='adjMatrix directory')
- parser.add_argument('--adjMatrix_reactom_dir', type=str, default='./prior_data_results/reactome_gene_interaction_adj_matrix_Sparse.npz', metavar='DIR', help='TF directory')
- parser.add_argument('--num_workers', type=int, default=2, metavar='N', help='the number of workers')
- parser.add_argument('--cell_type_key', type=str, default='cell_type', metavar='N', help='cell type key in scData.obs')
- parser.add_argument('--batch_size', type=int, default=3000, metavar='N', help='batch size for training(default: 128)')
- parser.add_argument('--epochs', type=int, default=10, metavar='N', help='number of epochs to train(default: 200)')
- parser.add_argument('--seed', type=int, default=1, metavar='S', help='random seed(default: 1)')
- parser.add_argument('--resume', type=str, default='', metavar='PATH', help='path to latest checkpoint(default: None)')
- parser.add_argument('--test_every', type=int, default=10, metavar='N', help='test after every epochs')
- parser.add_argument('--num_worker', type=int, default=1, metavar='N', help='the number of workers')
- parser.add_argument('--lr', type=float, default=1e-4, help='learning rate(default: 0.001)')
- parser.add_argument('--z_dim', type=int, default=100, metavar='N', help='the dim of latent variable z(default: 20)')
- parser.add_argument('--input_dim', type=int, default=1000, metavar='N')
- parser.add_argument('--place_holders', type=int, default=5, metavar='N', help='place holder number(default: 5)')
- parser.add_argument('--n_steps', type=int, default=200, metavar='N', help='diffusion steps(default: 200)')
- parser.add_argument('--sample_back_steps', type=int, default=30, metavar='N', help='diffusion sample_back_steps(default: 30)')
- parser.add_argument('--ddpm_pre_model_path', type=str, default='./%s/ddpm/model_best.pth'%args_ini.dataset_name, metavar='N', help='ddpm pre model saving directory')
- parser.add_argument('--ddpm_after_model_path', type=str, default='./%s/ddpm_after/model_best.pth'%args_ini.dataset_name, metavar='N', help='ddpm after model saving directory')
- args = parser.parse_args()
- kwargs = {'num_workers': 2, 'pin_memory': True} if cuda else {}
- os.makedirs('./%s/ddpm_after'%args_ini.dataset_name, exist_ok=True)
- def dataloader(batch_size=args.batch_size, num_workers=2):
- transform = transforms.Compose([
- transforms.ToTensor(),
- ])
- oriData = sc.read_h5ad(args.scData_dir)
- mtx_mask_0 = pd.read_csv(args.pathway_dir)
- mtx_mask_0 = mtx_mask_0.values
- print('pathway shape: ', mtx_mask_0.shape)
- mtx_mask_1 = pd.read_csv(args.complex_dir)
- mtx_mask_1 = mtx_mask_1.values
- print('complex shape: ', mtx_mask_1.shape)
- mtx_mask_2 = pd.read_csv(args.TF_dir)
- mtx_mask_2 = mtx_mask_2.values
- print('TF shape: ', mtx_mask_2.shape)
- mtx_mask = np.concatenate((mtx_mask_0,mtx_mask_1,mtx_mask_2),axis=1)
- print('final shape: ', mtx_mask.shape)
- sparse_adjmatrix_loaded = load_npz(args.adjMatrix_kegg_dir) #not symmetric
- gene_adj_mtx = sparse_adjmatrix_loaded.toarray()
- gene_adj_mtx = gene_adj_mtx + gene_adj_mtx.T
- gene_adj_mtx = gene_adj_mtx + np.eye(gene_adj_mtx.shape[0])
- reactom_adjmatrix_loaded = load_npz(args.adjMatrix_reactom_dir)
- reactom_gene_regu = reactom_adjmatrix_loaded.toarray()
- gene_adj_mtx = gene_adj_mtx + reactom_gene_regu
- gene_adj_mtx[gene_adj_mtx > 1] = 9
- ############################################
- n = args.place_holders
- if n != 0:
- vec = np.ones((mtx_mask.shape[0], n),dtype=int)
- mtx_mask = np.hstack((mtx_mask, vec))
- #gene_adj_mtx = np.hstack((gene_adj_mtx, vec))
- ############################################
- # fine-tunning data followed
- #####################################################
- fine_Data = sc.read_h5ad(args.fine_scData_dir)
- fine_gene = set(fine_Data.var_names.tolist())
- oriData_gene = set(oriData.var_names.tolist())
- intersect_gene = fine_gene.intersection(oriData_gene)
- gene_idx_ori = []
- gene_idx_fine = []
- for gene in intersect_gene:
- gene_idx_ori.append(oriData.var_names.tolist().index(gene))
- gene_idx_fine.append(fine_Data.var_names.tolist().index(gene))
- fineData_final = np.zeros((fine_Data.shape[0], oriData.shape[1]))
- data = fine_Data.X.toarray()
- for i in range(len(intersect_gene)):
- fineData_final[:, gene_idx_ori[i]] = data[:, gene_idx_fine[i]]
- #oriData_gene_num = oriData.shape[1]
- #fineData_gene_num = fine_Data.shape[1]
- #padding = np.zeros((fineData.shape[0], oriData_gene_num-fineData_gene_num),dtype=float)
- #fineData_final = np.hstack((data, padding))
- #####################################################
- cat = fineData_final
- cat = torch.Tensor(cat)
- label = fine_Data.obs[args.cell_type_key].tolist()
- label = [0 for x in label]
- lbls = torch.tensor(label)
- cat_Tdata = torch.utils.data.TensorDataset(cat, lbls)
- train_loader = torch.utils.data.DataLoader(dataset=cat_Tdata, batch_size=batch_size, shuffle=True)
- cat_test = fineData_final
- cat_test = torch.Tensor(cat_test)
- label_test = fine_Data.obs[args.cell_type_key].tolist()
- label_test = [0 for x in label_test]
- lbls_test = torch.tensor(label_test)
- cat_test_Tdata = torch.utils.data.TensorDataset(cat_test, lbls_test)
- test_loader = torch.utils.data.DataLoader(dataset=cat_test_Tdata, batch_size=batch_size, shuffle=True)
- classes = ('0', '1', '2', '3', '4', '5', '6', '7', '8', '9')
- return test_loader, train_loader, classes, mtx_mask, gene_adj_mtx
- def pdist(a,dim=2, p=2):
- dist_matrix = torch.norm(a[:, None]-a, dim, p)
- return dist_matrix
- def pdists(A, squared = False, eps = 1e-8):
- prod = torch.mm(A, A.t())
- norm = prod.diag().unsqueeze(1).expand_as(prod)
- res = (norm + norm.t() - 2 * prod).clamp(min = 0)
- if squared:
- return res
- else:
- res = res.clamp(min = eps).sqrt()
- return res
- def loss_function(x_hat, x, mu, log_var):
- """
- Calculate the loss. Note that the loss includes two parts.
- :param x_hat:
- :param x:
- :param mu:
- :param log_var:
- :return: total loss, BCE and KLD of our model
- """
- # 1. the reconstruction loss.
- # We regard the MNIST as binary classification
- x = x.view(x.shape[0],1,1,x.shape[-1])
- #BCE = F.binary_cross_entropy(x_hat, x, reduction='sum')
- BCE = F.mse_loss(x_hat, x, reduction='sum')
- # 2. KL-divergence
- # D_KL(Q(z|X) || P(z)); calculate in closed form as both dist. are Gaussian
- # here we assume that \Sigma is a diagonal matrix, so as to simplify the computation
- KLD = 0.5 * torch.sum(torch.exp(log_var) + torch.pow(mu, 2) - 1. - log_var)
- x_f = x.view(x.shape[0],x.shape[-1])
- graph_adj_loss = trustworthiness(x_f.cpu().detach().numpy() , mu.cpu().detach().numpy() , n_neighbors=25, metric='euclidean')
- graph_adj_loss = 1 - graph_adj_loss
- print('graph_adj_loss is: ',graph_adj_loss)
- loss = BCE + KLD +graph_adj_loss
- return loss, BCE, KLD
- def save_checkpoint(state, net, is_best, outdir):
- """
- 每训练一定的epochs后, 判断损失函数是否是目前最优的,并保存模型的参数
- :param state: 需要保存的参数,数据类型为dict
- :param is_best: 说明是否为目前最优的
- :param outdir: 保存文件夹
- :return:
- """
- if not os.path.exists(outdir):
- os.makedirs(outdir)
- checkpoint_file = os.path.join(outdir, 'checkpoint.pth')
- best_file = os.path.join(outdir, 'model_best.pth')
- torch.save(state, checkpoint_file)
- if is_best:
- shutil.copyfile(checkpoint_file, best_file)
- torch.save(net.state_dict(), args.ddpm_after_model_path)
- def test(model_vae, model_df, net, optimizer, mnist_test, epoch, best_test_loss):
- test_avg_loss = 0.0
- with torch.no_grad():
- for test_batch_index, (test_x, _) in enumerate(mnist_test):
- test_x = test_x.to(device)
- test_x_hat, test_mu, test_log_var = model_vae(test_x)
- test_loss, test_BCE, test_KLD = loss_function(test_x_hat, test_x, test_mu, test_log_var)
- img_shape = get_shape()
- current_batch_size = test_mu.shape[0]
- test_mu = test_mu.view(current_batch_size, 1, img_shape[1], img_shape[2])
- test_mu = test_mu.to(device)
- test_t = torch.randint(0, args.n_steps, (current_batch_size, )).to(device)
- eps = torch.randn_like(test_mu, dtype=torch.float).to(device)
- test_mu_t = model_df.sample_forward(test_mu, test_t, eps)
- test_mu_t = test_mu_t.to(device)
- eps_theta = net(test_mu_t, test_t.reshape(current_batch_size, 1))
- #loss_fn = nn.MSELoss(reduction='sum')
- loss_fn = nn.MSELoss()
- loss_df = loss_fn(eps_theta, eps)
- test_loss = test_loss + loss_df
- test_avg_loss += test_loss
- test_avg_loss /= len(mnist_test.dataset)
- is_best = test_avg_loss < best_test_loss
- best_test_loss = min(test_avg_loss, best_test_loss)
- save_checkpoint({
- 'epoch': epoch,
- 'best_test_loss': best_test_loss,
- 'state_dict': model_vae.state_dict(),
- 'optimizer': optimizer.state_dict(),
- }, net, is_best, args.vae_after_dir)
- return best_test_loss
- def get_shape():
- return (1, 10, 10)
- def main():
- mnist_test, mnist_train, classes, mtx_mask, gene_adj_mtx= dataloader(args.batch_size, args.num_worker)
- x, label = iter(mnist_train).__next__()
- print("mask_mtx.size = ", mtx_mask.shape)
- print("pathway_connect_sum = ", sum(mtx_mask))
- print("Max(pathway_connect_sum) = ", max(sum(mtx_mask)))
- print("Min(pathway_connect_sum) = ", min(sum(mtx_mask)))
- print("mask_sum = ", np.sum(mtx_mask))
- print("zero_genes_pathway = ", np.sum(sum(mtx_mask)==0))
- print("zero_pathway_genes = ", np.sum(np.sum(mtx_mask,axis=1)==0))
- model_vae = VAE(mask_mtx=mtx_mask, gene_adj_mtx=gene_adj_mtx,z_dim=args.z_dim)
- model_vae.load_state_dict(torch.load("%s/model_best.pth"%args.vae_pre_dir)['state_dict'])
- model_vae = model_vae.to(device)
- print('The structure of our model is shown below: \n')
- print(model_vae)
- convnet_small_cfg = {
- 'type': 'ConvNet',
- 'intermediate_channels': [10, 20],
- 'pe_dim': 128
- }
- convnet_medium_cfg = {
- 'type': 'ConvNet',
- 'intermediate_channels': [10, 10, 20, 20, 40, 40, 80, 80],
- 'pe_dim': 256,
- 'insert_t_to_all_layers': True
- }
- convnet_big_cfg = {
- 'type': 'ConvNet',
- 'intermediate_channels': [20, 20, 40, 40, 80, 80, 160, 160],
- 'pe_dim': 256,
- 'insert_t_to_all_layers': True
- }
- unet_1_cfg = {'type': 'UNet',
- 'channels': [10, 20, 40, 80],
- 'pe_dim': 128}
- unet_res_cfg = {
- 'type': 'UNet',
- 'channels': [10, 20, 40, 80],
- 'pe_dim': 128,
- 'residual': True
- }
- configs = [convnet_small_cfg, convnet_medium_cfg, convnet_big_cfg, unet_1_cfg, unet_res_cfg]
- n_epochs = args.epochs
- n_steps = args.n_steps
- sample_back_steps = args.sample_back_steps
- config_id = 4
- config = configs[config_id]
- net = build_network(config, n_steps)
- net.load_state_dict(torch.load(args.ddpm_pre_model_path))
- net = net.to(device)
- model_df = DDPM(device, n_steps, sample_back_steps)
- #optimizer = optim.Adam(model_vae.parameters(), lr=args.lr)
- optimizer = optim.Adam(list(model_vae.parameters()) + list(net.parameters()), lr=args.lr)
- # Step 3: optionally resume from a checkpoint
- start_epoch = 0
- best_test_loss = np.finfo('f').max
- if args.resume:
- if os.path.isfile(args.resume):
- print('=> loading checkpoint %s' % args.resume)
- checkpoint = torch.load(args.resume)
- start_epoch = checkpoint['epoch'] + 1
- best_test_loss = checkpoint['best_test_loss']
- model.load_state_dict(checkpoint['state_dict'])
- optimizer.load_state_dict(checkpoint['optimizer'])
- print('=> loaded checkpoint %s' % args.resume)
- else:
- print('=> no checkpoint found at %s' % args.resume)
- if not os.path.exists(args.vae_after_result_dir):
- os.makedirs(args.vae_after_result_dir)
- loss_epoch = []
- for epoch in range(start_epoch, args.epochs):
- time_start = time.time()
- loss_batch = []
- for batch_index, (x, _) in enumerate(mnist_train):
- #x = x.view(x.shape[0],1,1,x.shape[1])
- x = x.to(device)
- x_hat, mu, log_var = model_vae(x)
- loss_vae, BCE, KLD = loss_function(x_hat, x, mu, log_var)
- img_shape = get_shape()
- current_batch_size = mu.shape[0]
- mu = mu.view(current_batch_size, 1, img_shape[1], img_shape[2])
- mu = mu.to(device)
- t = torch.randint(0, args.n_steps, (current_batch_size, )).to(device)
- eps = torch.randn_like(mu).to(device)
- mu_t = model_df.sample_forward(mu, t, eps)
- mu_t = mu_t.to(device)
- #print("mu_t is on:", mu_t.device)
- #print("t is on:", t.device)
- eps_theta = net(mu_t, t.reshape(current_batch_size, 1))
- loss_fn = nn.MSELoss()
- loss_df = loss_fn(eps_theta, eps) * current_batch_size
- loss = loss_vae + loss_df
- loss_batch.append(loss_vae.item() + loss_df.item())
- optimizer.zero_grad()
- loss.backward()
- optimizer.step()
- # print statistics every 100 batch
- if (batch_index + 1) % 10 == 0:
- print('Epoch [{}/{}], Batch [{}/{}] : Total-loss = {:.4f}, BCE-Loss = {:.4f}, KLD-loss = {:.4f}, DDPM-loss = {:.4f}'
- .format(epoch + 1, args.epochs, batch_index + 1, len(mnist_train.dataset) // args.batch_size,
- loss.item() / args.batch_size, BCE.item() / args.batch_size,
- KLD.item() / args.batch_size, loss_df.item() / args.batch_size))
- loss_epoch.append(np.sum(loss_batch) / len(mnist_train.dataset))
- time_end = time.time()
- time_used = time_end - time_start
- print("VAE epoch %s used time: %s sec."%(str(epoch),format(time_end-time_start, '.5f')))
- if (epoch + 1) % args.test_every == 0:
- best_test_loss = test(model_vae,model_df,net, optimizer, mnist_test, epoch, best_test_loss)
- return loss_epoch
- if __name__ == '__main__':
- loss_epoch = main()
- #print(loss_epoch)
- save_file(args.dataset_name, args.ddpm_after_model_path,args.batch_size,args.num_workers, args.scData_dir,args.fine_scData_dir, args.pathway_dir,args.complex_dir,args.TF_dir,args.adjMatrix_kegg_dir,args.adjMatrix_reactom_dir,args.place_holders, args.vae_after_dir, args.n_steps, args.sample_back_steps)
main_joint_fine_tunning.py at commit ff4ab8e, no license · at the source
Overview
Abstract
Single-cell RNA sequencing data rely heavily on dimensionality reduction methods to uncover underlying structures and patterns. Nonlinear dimensionality-reduction
Reproduced under the paper's license (CC BY-NC), from the paper cited above.
Repositories
Its files are read in the Code ↔ Paper reader above, with 6 matches between paragraphs and lines of code.
MaLab-scGenomics/PriorVAED
ff4ab8e2273c04635f5075cfe5ac79af17d4e775, 4 November 2025Availability: 1 check, the latest on 27 September 2026: the link answers
- 27 September 2026: the link answers
19 files
- PVAED/
DDPM.py , Python, 522 lines - PVAED/
customized_linear.py , Python, 150 lines - PVAED/
data_process.ipynb , Jupyter, 240 lines - PVAED/
diffusion.py , Python, 507 lines - PVAED/
docs/ , JavaScript, 149 lines_build/ html/ _static/ doctools.js - PVAED/
docs/ , JavaScript, 13 lines_build/ html/ _static/ documentation_options.js - PVAED/
docs/ , JavaScript, 192 lines_build/ html/ _static/ language_data.js - PVAED/
docs/ , JavaScript, 635 lines_build/ html/ _static/ searchtools.js - PVAED/
docs/ , JavaScript, 154 lines_build/ html/ _static/ sphinx_highlight.js - PVAED/
docs/ , JavaScript, 1 line_build/ html/ searchindex.js - PVAED/
docs/ , Python, 28 linesconf.py - PVAED/
docs/ , Jupyter, 51 linesdown_stream.ipynb - PVAED/
down_stream.ipynb , Jupyter, 51 lines, 1 match - PVAED/
main.py , Python, 321 lines, 2 matches - PVAED/
main_joint_fine_tunning. , Python, 387 lines, 3 matchespy - PVAED/
utils.py , Python, 109 lines - PVAED/
utils_joint.py , Python, 211 lines - PVAED/
vae.py , Python, 100 lines - README.md, Text, 18 lines
pvaed.readthedocs.io
Availability: 1 check, the latest on 27 September 2026: the link answers (HTTP 200)
- 27 September 2026: the link answers (HTTP 200)
Code availability
The current version of source code and software implementation for PVAED is available at: https://
Reproduced under the paper's license (CC BY-NC), from the paper cited above.
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;
- 18 scripts, each with its path and the digest of its content;
- 6 matches 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
PBMC 3k and 10k data are downloaded from 10x genomics website under the corresponding keywords at https://
Reproduced under the paper's license (CC BY-NC), 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 1, 27 September 2026: the first record
Recorded: type, language, journal, volume, issue, pages, dates, 3 authors, 5 keywords, 9 MeSH terms, 49 references.
Cite
This paper
Niu, Y., Chen, Y., & Ma, W. (2026). PVAED: prior-guided variational autoencoders with diffusion denoising for interpretable single-cell representation learning. Briefings in bioinformatics, 27(3), bbag259. https://
BibTeX
@article{niu2026pvaed,
author = {Niu, Yawei and Chen, Yichu and Ma, Wenji},
title = {{PVAED: prior-guided variational autoencoders with diffusion denoising for interpretable single-cell representation learning}},
journal = {Briefings in bioinformatics},
year = {2026},
month = may,
volume = {27},
number = {3},
pages = {bbag259},
publisher = {Oxford University Press},
issn = {1467-5463},
doi = {10.1093/
url = {https://
pmid = {42202286},
pmcid = {PMC13215596}
}
RIS
TY - JOUR
AU - Niu, Yawei
AU - Chen, Yichu
AU - Ma, Wenji
TI - PVAED: prior-guided variational autoencoders with diffusion denoising for interpretable single-cell representation learning
T2 - Briefings in bioinformatics
J2 - Brief Bioinform
PY - 2026
DA - 2026/
VL - 27
IS - 3
SP - bbag259
SN - 1467-5463
PB - Oxford University Press
DO - 10.1093/
UR - https://
LA - en
ER -
CSL-JSON
{
"id": "10.1093/
"type": "article-journal",
"title": "PVAED: prior-guided variational autoencoders with diffusion denoising for interpretable single-cell representation learning",
"container-title": "Briefings in bioinformatics",
"author": [
{
"family": "Niu",
"given": "Yawei"
},
{
"family": "Chen",
"given": "Yichu"
},
{
"family": "Ma",
"given": "Wenji"
}
],
"container-title-short":
"volume": "27",
"issue": "3",
"page": "bbag259",
"DOI": "10.1093/
"PMID": "42202286",
"PMCID": "PMC13215596",
"ISSN": "1467-5463",
"publisher": "Oxford University Press",
"URL": "https://
"language": "en",
"issued": {
"date-parts": [
[
2026,
5,
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:10.1186/s13059-023-02850-y [code]
- siVAE: interpretable deep generative models for single-cell transcriptomes.Journal: Genome biologyIn common: anndata, Scanpy, PyTorch, 5 other tools, genetics / omics, 9 references
- [2] doi:10.1016/j.xgen.2024.100739 [code]
- CSI-GEP: A GPU-based unsupervised machine learning approach for recovering gene expression programs in atlas-scale single-cell RNA-seq data.Journal: Cell genomicsIn common: anndata, Scanpy, NetworkX, 6 other tools, genetics / omics, 6 references
- [3] doi:10.1038/s41467-021-25534-2 [code]
- Learning interpretable cellular and gene signature embeddings from single-cell transcriptomic data.Journal: Nature communicationsIn common: anndata, Scanpy, PyTorch, 5 other tools, genetics / omics, 6 references
- [4] doi:10.1093/nar/gkag706 [code]
- scDifformer: diffusion-based post-training for virtual cell modeling across large-scale single-cell data.Journal: Nucleic acids researchIn common: anndata, Scanpy, NetworkX, 6 other tools, 4 references
- [5] doi:10.1186/s12864-026-12965-8 [code]
- Systematic evaluation of single-cell foundation model interpretability: attention-derived edge scores add no incremental value over gene-level features for perturbation-target prediction.Journal: BMC genomicsIn common: anndata, Scanpy, NetworkX, 6 other tools, 4 references
- [6] doi:10.1038/s41588-025-02120-6 [code]
- Quantitative characterization of cell niches in spatially resolved omics data.Journal: Nature geneticsIn common: anndata, Scanpy, NetworkX, 6 other tools, genetics / omics, 4 references
- [7] doi:10.1038/s41467-026-71759-4 [code]
- CellNiche represents cellular microenvironments in atlas-scale spatial omics data with contrastive learning.Journal: Nature communicationsIn common: anndata, Scanpy, NetworkX, 6 other tools, 4 references
- [8] doi:10.1038/s41467-024-53971-2 [code]
- scPair: Boosting single cell multimodal analysis by leveraging implicit feature selection and single cell atlases.Journal: Nature communicationsIn common: anndata, Scanpy, PyTorch, 5 other tools, genetics / omics, 4 references
- [9] doi:10.1038/s42003-025-07872-9 [code]
- Infusing structural assumptions into dimensionality reduction for single-cell RNA sequencing data to identify small gene sets.Journal: Communications biologyIn common: anndata, Scanpy, scikit-learn, 4 other tools, genetics / omics, 4 references
- [10] doi:10.1101/gr.280803.125 [code]
- Unified integration of spatial transcriptomics across platforms with LLOKI.Journal: Genome researchIn common: anndata, Scanpy, PyTorch, 4 other tools, genetics / omics, 4 references
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, 18 scripts, and 6 matches 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: sha256:85fb10e1922928f0…
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/10.1093/bib/bbag259)
<!-- oscr-badge:end -->reStructuredText
.. oscr-badge:start
.. image:: https://openscicode.org/badge/doi/10.1093/bib/bbag259.svg
:target: https://openscicode.org/badge/doi/10.1093/bib/bbag259
:alt: OSCR
.. oscr-badge:endHTML
<!-- oscr-badge:start -->
<a href="https://openscicode.org/badge/doi/10.1093/bib/bbag259"><img src="https://openscicode.org/badge/doi/10.1093/bib/bbag259.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".
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.
