OSCR

Empowering multifaceted analysis of spatial transcriptomics data with RGAST.

Code ↔ Paper

11 matches between paragraphs of the paper and lines of its authors' code, computed by the harvester (lexical-v1). Click a colored paragraph or line to see its counterpart.

The 11 matches
  1. [1] § Materials and methods › Spatially variable genes detection › Detection of SVGs ↔ tutorial/SVG_detection.ipynb, lines 76–114 · score 0.70 · fold change, neighboring domain, fraction, ratio, DE, rank
  2. [2] § Materials and methods › Cell–cell communications analysis ↔ RGAST/cci.py, lines 23–72 · score 0.68 · CellChatDB, LR pair, database, subunits, ligand, receptor
  3. [3] § Materials and methods › Cell–cell communications analysis ↔ RGAST/cci.py, lines 418–476 · score 0.65 · spatial vector fields, signaling direction, weights, matrix, network, spot
  4. [4] § Materials and methods › Trajectory and pseudotime inference › Cross-sectional 3D RGAST model ↔ tutorial/3D_RGAST.ipynb, lines 1–23 · score 0.65 · discontinuous independent technical, spatial neighborhoods lies, consecutive, noises, adjacent, biological
  5. [5] § Materials and methods › Deep embedding clustering (optional) ↔ RGAST/Train_RGAST.py, lines 255–378 · score 0.63 · deep embedded clustering, RGAST embedding, DEC, latent
  6. [6] § Materials and methods › Relational graph attention auto-encoder › Loss function ↔ RGAST/Train_RGAST.py, lines 77–118 · score 0.63 · Adam optimizer, weight decay, Loss
  7. [7] § Results › Overview of the RGAST model ↔ RGAST/Train_RGAST.py, lines 77–118 · score 0.62 · graph attention auto, latent embedding, dimensional, profiles, encoder, reconstructed
  8. [8] § Materials and methods › Scalable training of RGAST ↔ tutorial/train_with_DIC.ipynb, lines 1–21 · score 0.61 · mouse olfactory bulb, DIC strategy, Stereo seq, Training
  9. [9] § Materials and methods › Scalable training of RGAST ↔ RGAST/Train_RGAST.py, lines 25–74 · score 0.54 · Training RGAST, spatial transcriptomics, batch, networks
  10. [10] § Results › Validation and application of RGAST in deciphering spatially resolved cell–cell communication networks ↔ tutorial/de_novo_CCC_analysis.ipynb, lines 148–162 · score 0.53 · Penk Oprd1, ligand receptor, CCC, score, pathway, cell
  11. [11] § Results › Validation and application of RGAST in deciphering spatially resolved cell–cell communication networks ↔ tutorial/de_novo_CCC_analysis.ipynb, lines 148–162 · score 0.52 · ligand receptor pair, Penk Oprd1, CCC, scores, pathways, cell

Paper

Loaded from Europe PMC by your browser, not stored by OSCR: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The authors' code

Python · 394 lines · 18 KB · MIT · 4 matches

  1. import numpy as np
  2. import os
  3. import scanpy as sc
  4. import anndata
  5. from sklearn.metrics.cluster import adjusted_rand_score
  6. from sklearn.metrics import silhouette_score
  7. import random
  8. from tqdm import tqdm
  9. import warnings
  10. from .RGAST import RGAST
  11. from .utils import Transfer_pytorch_Data, res_search_fixed_clus, Batch_Data, Cal_Spatial_Net, Cal_Expression_Net, mclust_R
  13. import torch
  14. import torch.backends.cudnn as cudnn
  15. cudnn.benchmark = True
  16. import torch.nn.functional as F
  17. from torch_geometric.loader import DataLoader
  19. def target_distribution(batch):
  20. weight = (batch ** 2) / torch.sum(batch, 0)
  21. return (weight.t() / torch.sum(weight, 1)).t()
  23. class Train_RGAST:
  25. def __init__(self, adata, dim_reduction = None, batch_data = False, num_batch_x_y = None, device_idx = 7, spatial_net_arg = {}, exp_net_arg = {}, verbose=True, center_msg='out'):
  27. """\
  28. Initialization of a RGAST trainer.
  30. Parameters
  31. ----------
  32. adata
  33. AnnData object of scanpy package.
  34. num_batch_x_y
  35. A tuple specifying the number of points at which to segment the spatially transcribed image on the x and y axes.
  36. Each split is then trained as a batch. This is useful for large scale cases.
  37. spatial_net_arg
  38. A dict passing key-word arguments to calculating spatial network in each batch data. See `Cal_Spatial_Net`.
  39. exp_net_arg
  40. A dict passing key-word arguments to calculating expression network in each batch data. See `Cal_Expression_Net`
  41. """
  43. if dim_reduction == 'PCA':
  44. if 'X_pca' not in adata.obsm.keys():
  45. raise ValueError("PCA has not been done! Run sc.pp.pca first!")
  46. elif dim_reduction == 'HVG':
  47. if 'highly_variable' not in adata.var.keys():
  48. raise ValueError("HVG has not been computed! Run sc.pp.highly_variable_genes first!")
  49. else:
  50. warnings.warn("No dimentional reduction method specified, using all genes' expression as input.")
  52. self.dim_reduction = dim_reduction
  53. self.batch_data = batch_data
  54. self.adata = adata
  56. if 'Spatial_Net' not in adata.uns.keys():
  57. raise ValueError("Spatial_Net is not existed! Run Cal_Spatial_Net first!")
  58. if 'Exp_Net' not in adata.uns.keys():
  59. raise ValueError("Exp_Net is not existed! Run Cal_Expression_Net first!")
  60. self.data = Transfer_pytorch_Data(adata, dim_reduction=dim_reduction,center_msg=center_msg)
  61. if verbose:
  62. print('Size of Input: ', self.data.x.shape)
  64. if batch_data:
  65. self.num_batch_x, self.num_batch_y = num_batch_x_y
  66. Batch_list = Batch_Data(adata, num_batch_x=self.num_batch_x, num_batch_y=self.num_batch_y)
  67. for temp_adata in Batch_list:
  68. Cal_Spatial_Net(temp_adata, **spatial_net_arg)
  69. Cal_Expression_Net(temp_adata, dim_reduce=dim_reduction, **exp_net_arg)
  70. data_list = [Transfer_pytorch_Data(adata, dim_reduction=dim_reduction,center_msg=center_msg) for adata in Batch_list]
  71. self.loader = DataLoader(data_list, batch_size=1, shuffle=True)
  73. self.device = torch.device(f'cuda:{device_idx}' if torch.cuda.is_available() else 'cpu')
  74. self.model = None
  77. def train_RGAST(self, early_stopping = True, label_key = None, save_path = '.', n_clusters = 7, cluster_method = 'leiden',
  78. hidden_dims=[100, 32], n_epochs=1000, lr=0.001, key_added='RGAST', att_drop = 0.3,
  79. gradient_clipping=5., weight_decay=0.0001, min_epochs=300, random_seed=0, save_loss=False,
  80. save_reconstrction=False, save_attention=True):
  82. """\
  83. Training graph attention auto-encoder.
  85. Parameters
  86. ----------
  87. early_stopping
  88. Using early stopping strategy or not. Default = True.
  89. lable_key
  90. A key specify the specific column in adata.obs to be treated as reference label.
  91. save_path
  92. directory to save the trained RGAST model.
  93. n_clusters
  94. number of clusters to set when calculating early stopping criterion.
  95. hidden_dims
  96. The dimension of the encoder (depends on RGAST or RGAST2).
  97. n_epochs
  98. Number of total epochs in training.
  99. lr
  100. Learning rate for AdamOptimizer.
  101. key_added
  102. The latent embeddings are saved in adata.obsm[key_added].
  103. gradient_clipping
  104. Gradient Clipping.
  105. weight_decay
  106. Weight decay for AdamOptimizer.
  107. save_loss
  108. If True, the training loss is saved in adata.uns['RGAST_loss'].
  109. save_reconstrction
  110. If True, the reconstructed PCA profiles are saved in adata.layers['RGAST_ReX'].
  112. Returns
  113. -------
  114. AnnData
  115. """
  116. self.save_path = save_path
  117. self.label_key = label_key
  118. self.n_clusters = n_clusters
  120. # seed_everything()
  121. seed=random_seed
  122. random.seed(seed)
  123. np.random.seed(seed)
  124. torch.manual_seed(seed)
  125. torch.cuda.manual_seed(seed)
  128. if self.model is None:
  129. model = RGAST(hidden_dims = [self.data.x.shape[1]] + hidden_dims, dim_reduce=self.dim_reduction, att_drop=att_drop).to(self.device)
  130. else:
  131. model = self.model.to(self.device)
  133. data = self.data.to(self.device)
  135. optimizer = torch.optim.Adam(model.parameters(), lr=lr, weight_decay=weight_decay)
  137. loss_list = []
  138. score_list = [0]
  139. num_fail = 0
  141. with tqdm(range(n_epochs)) as tq:
  142. for epoch in tq:
  143. if early_stopping:
  144. with torch.no_grad():
  145. if label_key is not None:
  146. if (epoch+1) % 50 == 0:
  147. if self.batch_data:
  148. model.to('cpu')
  149. model.eval()
  150. z, _, _, _ = model(data.x.cpu(), data.edge_index.cpu(), data.edge_type.cpu())
  151. model.to(self.device)
  152. else:
  153. model.eval()
  154. z, _, _, _ = model(data.x, data.edge_index, data.edge_type)
  155. z = z.to('cpu').detach().numpy()
  156. adata_RGAST = anndata.AnnData(z)
  157. adata_RGAST.obs_names=self.adata.obs_names
  158. if cluster_method == 'mclust':
  159. mclust_R(adata_RGAST, n_clusters)
  160. elif cluster_method == 'leiden':
  161. sc.pp.neighbors(adata_RGAST)
  162. _ = res_search_fixed_clus(adata_RGAST, n_clusters)
  163. obs_df = adata_RGAST.obs.join(self.adata.obs[label_key]).dropna(subset=label_key)
  164. ARI = adjusted_rand_score(obs_df['RGAST'], obs_df[label_key])
  165. if ARI <= max(score_list):
  166. num_fail += 1
  167. if num_fail>3 and epoch>=min_epochs:
  168. break
  169. else:
  170. num_fail = 0
  171. torch.save(model,f'{save_path}/model.pth')
  172. self.adata.obs['RGAST'] = adata_RGAST.obs['RGAST']
  173. score_list.append(ARI)
  174. tq.set_postfix(ARI=round(max(score_list),3))
  176. else:
  177. if (epoch+1) % 50 == 0:
  178. if self.batch_data:
  179. model.to('cpu')
  180. model.eval()
  181. z, _, _, _ = model(data.x.cpu(), data.edge_index.cpu(), data.edge_type.cpu())
  182. model.to(self.device)
  183. else:
  184. model.eval()
  185. z, _, _, _ = model(data.x, data.edge_index, data.edge_type)
  186. z = z.to('cpu').detach().numpy()
  187. adata_RGAST = anndata.AnnData(z)
  188. adata_RGAST.obs_names=self.adata.obs_names
  189. sc.pp.neighbors(adata_RGAST)
  190. _ = res_search_fixed_clus(adata_RGAST, n_clusters)
  191. SC = silhouette_score(z, adata_RGAST.obs['RGAST'])
  192. if SC <= max(score_list):
  193. num_fail += 1
  194. if num_fail>3 and epoch>=min_epochs:
  195. break
  196. else:
  197. num_fail = 0
  198. torch.save(model,f'{save_path}/model.pth')
  199. self.adata.obs['RGAST'] = adata_RGAST.obs['RGAST']
  200. score_list.append(SC)
  201. tq.set_postfix(SC=round(max(score_list),3))
  203. if self.batch_data:
  204. for batch in self.loader:
  205. batch = batch.to(self.device)
  206. model.train()
  207. optimizer.zero_grad()
  208. z, out, _, _ = model(batch.x, batch.edge_index, batch.edge_type)
  209. loss = F.mse_loss(batch.x, out) #F.nll_loss(out[data.train_mask], data.y[data.train_mask])
  210. loss.backward()
  211. loss_list.append(loss.item())
  212. torch.nn.utils.clip_grad_norm_(model.parameters(), gradient_clipping)
  213. optimizer.step()
  215. else:
  216. model.train()
  217. optimizer.zero_grad()
  218. z, out, _, _ = model(data.x, data.edge_index, data.edge_type)
  219. loss = F.mse_loss(data.x, out) #F.nll_loss(out[data.train_mask], data.y[data.train_mask])
  220. loss.backward()
  221. loss_list.append(loss.item())
  222. torch.nn.utils.clip_grad_norm_(model.parameters(), gradient_clipping)
  223. optimizer.step()
  225. if early_stopping == True and os.path.exists(f'{save_path}/model.pth'):
  226. model = torch.load(f'{save_path}/model.pth',weights_only=False).to(self.device)
  228. with torch.no_grad():
  229. if self.batch_data:
  230. model.to('cpu')
  231. model.eval()
  232. z, out, att1, att2 = model(data.x.cpu(), data.edge_index.cpu(), data.edge_type.cpu())
  233. model.to(self.device)
  234. else:
  235. model.eval()
  236. z, out, att1, att2 = model(data.x, data.edge_index, data.edge_type)
  238. RGAST_rep = z.to('cpu').detach().numpy()
  239. np.save(f'{save_path}/RGAST_embedding.npy', RGAST_rep)
  240. torch.save(model.to('cpu'),f'{save_path}/model.pth')
  241. self.adata.obsm[key_added] = RGAST_rep
  243. if save_loss:
  244. self.adata.uns['RGAST_loss'] = loss_list
  245. if save_reconstrction:
  246. ReX = out.to('cpu').numpy()
  247. self.adata.obsm['RGAST_ReX'] = ReX
  248. if save_attention:
  249. self.adata.uns['att1'] = (att1[0].to('cpu').numpy(),att1[1].to('cpu').numpy())
  250. self.adata.uns['att2'] = (att2[0].to('cpu').numpy(),att2[1].to('cpu').numpy())
  252. self.model = model
  255. def train_with_dec(self, verbose = True, early_stopping = True, key_added='RGAST', num_epochs=1000, dec_interval=50, dec_tol=0.01):
  257. """\
  258. Training graph attention auto-encoder with deep embedding clustering.
  259. Only call this after call Train_RGAST.train_RGAST() and make sure batch_data = False.
  261. Parameters
  262. ----------
  263. early_stopping
  264. Using early stopping strategy or not. Default = True.
  265. key_added
  266. The latent embeddings are saved in adata.obsm[key_added].
  267. num_epochs
  268. Number of total epochs in training.
  269. dec_interval
  270. Evaluate after how many epochs (for early stopping).
  271. dec_tol
  272. DEC tol.
  274. Returns
  275. -------
  276. AnnData with updated .obsm[key_added]
  277. """
  279. # initialize cluster parameter
  280. model = self.model.to(self.device)
  281. model.eval()
  282. test_z = self.adata.obsm['RGAST']
  283. y_pred_last = np.array(self.adata.obs['RGAST'],dtype=np.int32).copy()
  284. counts = len(np.bincount(y_pred_last))
  285. cluster_layer = []
  286. for i in range(counts):
  287. cluster_layer.append(np.mean(test_z[y_pred_last==i,],axis=0))
  288. cluster_layer = torch.tensor(cluster_layer).to(self.device)
  289. data = self.data.to(self.device)
  291. optimizer = torch.optim.Adam(model.parameters(), lr=0.001, weight_decay=0.0001)
  293. score_list = [0]
  294. num_fail = 0
  296. with tqdm(range(num_epochs)) as tq:
  297. for epoch_id in tq:
  299. if (epoch_id+1) % dec_interval == 0:
  301. if early_stopping:
  302. #early stopping
  303. if self.label_key is not None:
  304. model.eval()
  305. z, _ = model(data.x, data.edge_index, data.edge_type)
  306. z = z.to('cpu').detach().numpy()
  307. adata_RGAST = anndata.AnnData(z)
  308. adata_RGAST.obs_names=self.adata.obs_names
  309. sc.pp.neighbors(adata_RGAST)
  310. _ = res_search_fixed_clus(adata_RGAST, self.n_clusters)
  311. obs_df = adata_RGAST.obs.join(self.adata.obs[self.label_key]).dropna(subset=self.label_key)
  312. ARI = adjusted_rand_score(obs_df['RGAST'], obs_df[self.label_key])
  313. if ARI <= max(score_list):
  314. num_fail += 1
  315. if num_fail>3 and epoch_id>=300:
  316. break
  317. else:
  318. num_fail = 0
  319. torch.save(model,f'{self.save_path}/model.pth')
  320. self.adata.obs['RGAST'] = adata_RGAST.obs['RGAST']
  321. score_list.append(ARI)
  322. tq.set_postfix(ARI=round(max(score_list),3))
  324. else:
  325. model.eval()
  326. z, _ = model(data.x, data.edge_index, data.edge_type)
  327. z = z.to('cpu').detach().numpy()
  328. adata_RGAST = anndata.AnnData(z)
  329. adata_RGAST.obs_names=self.adata.obs_names
  330. sc.pp.neighbors(adata_RGAST)
  331. _ = res_search_fixed_clus(adata_RGAST, self.n_clusters)
  332. SC = silhouette_score(z, adata_RGAST.obs['RGAST'])
  333. if SC <= max(score_list):
  334. num_fail += 1
  335. if num_fail>3 and epoch_id>=300:
  336. break
  337. else:
  338. num_fail = 0
  339. torch.save(model,f'{self.save_path}/model.pth')
  340. self.adata.obs['RGAST'] = adata_RGAST.obs['RGAST']
  341. score_list.append(SC)
  342. tq.set_postfix(SC=round(max(score_list),3))
  344. #DEC update
  345. z, reconst = model(data.x, data.edge_index, data.edge_type)
  346. q = 1.0 / (1.0 + torch.sum(torch.pow(z.unsqueeze(1) - cluster_layer, 2), 2))
  347. q = (q.t() / torch.sum(q, 1)).t()
  348. tmp_p = target_distribution(torch.Tensor(q))
  349. y_pred = tmp_p.cpu().detach().numpy().argmax(1)
  350. delta_label = np.sum(y_pred != y_pred_last).astype(np.float32) / y_pred.shape[0]
  351. y_pred_last = np.copy(y_pred)
  352. if epoch_id > 0 and delta_label < dec_tol:
  353. print('delta_label {:.4}'.format(delta_label), '< tol', dec_tol)
  354. print('Reached tolerance threshold. Stopping training.')
  355. break
  358. # training model
  359. model.train()
  360. optimizer.zero_grad()
  361. z, reconst = model(data.x, data.edge_index, data.edge_type)
  362. q = 1.0 / (1.0 + torch.sum(torch.pow(z.unsqueeze(1) - cluster_layer, 2), 2) / 1.0)
  363. q = (q.t() / torch.sum(q, 1)).t()
  364. loss_rec = F.mse_loss(data.x, reconst)
  365. # clustering KL loss
  366. loss_kl = F.kl_div(q.log(), torch.tensor(tmp_p).to(self.device)).to(self.device)
  367. loss = loss_kl + loss_rec
  368. loss.backward()
  369. optimizer.step()
  371. model = torch.load(f'{self.save_path}/model.pth').to(self.device)
  372. model.eval()
  373. z, _ = model(data.x, data.edge_index, data.edge_type)
  375. RGAST_rep = z.to('cpu').detach().numpy()
  376. np.save(f'{self.save_path}/RGAST_embedding.npy', RGAST_rep)
  377. self.adata.obsm[key_added] = RGAST_rep
  378. self.model = model
  380. def load_model(self, path):
  381. self.model = torch.load(path, map_location=self.device)
  383. def save_model(self, path):
  384. torch.save(self.model,f'{path}/model.pth')
  386. @torch.no_grad()
  387. def process(self, gdata = None):
  388. if gdata is None:
  389. gdata = self.data
  390. self.model.to(self.device)
  391. self.model.eval()
  392. gdata = gdata.to(self.device)
  393. return self.model(gdata.x, gdata.edge_index, gdata.edge_type)

Train_RGAST.py at commit efea2dd, under MIT · at the source

Overview

Authors: Yuqiao Gong1, Xin Yuan1,2,3,4, Zhangsheng Yu1,2,3,5
ORCID iDs: Yuqiao Gong, Xin Yuan, Zhangsheng Yu
  1. Department of Bioinformatics and Biostatistics, School of Life Sciences and Biotechnology, Shanghai Jiao Tong University, 800 Dongchuan Road, Minhang District, Shanghai, Shanghai, 200240, China
  2. SJTU-Yale Joint Center for Biostatistics and Data Science Organization, Shanghai Jiao Tong University, 800 Dongchuan Road, Minhang District, Shanghai, Shanghai, 200240, China
  3. Institute of Translational Medicine, Shanghai Jiao Tong University, 800 Dongchuan Road, Minhang District, Shanghai, Shanghai, 200240, China
  4. National Center for Translational Medicine, Shanghai Jiao Tong University, 800 Dongchuan Road, Minhang District, Shanghai, Shanghai, 200240, China
  5. Institute of Clinical Medicine, School of Medicine, Shanghai Jiao Tong University, 800 Dongchuan Road, Minhang District, Shanghai, Shanghai, 200240, China
Institutions: Shanghai Jiao Tong University (China)
Journal: Briefings in bioinformatics, volume 27, issue 3, article bbag298
Dates: received 20 October 2025; accepted 15 May 2026; published online 16 June 2026; in print May 2026
Type: Research article · Language: English
License: CC BY
Identifiers: DOI 10.1093/bib/bbag298 · PMID 42302280 · PMCID PMC13271401 · OpenAlex W7164902072
Open access: gold, a free copy (OpenAlex)
Status: code verified
Categories: genetics / omics (modality), human (organism), mouse (organism), methods / tools (subfield)
Methods: Statistics, Smoothing, state filtering, decompositions, Machine learning
Keywords: spatial transcriptomics, heterogeneous graph network, relational graph attention, cell–cell communication, RGAST
MeSH: Software*, Spatial Transcriptomics*, Algorithms, Animals, Computational Biology, Humans, Mice (* major topic)
Topic: Single-cell and spatial transcriptomics (Molecular Biology, Biochemistry, Genetics and Molecular Biology), according to OpenAlex
Funding: Shanghai Key Laboratory of Child Brain and Development (24dz2260100); Shanghai Jiao Tong University - Luoyang Orthopedic-Traumatological Hospital of Henan Province Collaborative Research Fund (ZZSW-HLOH-2025006); National Natural Science Foundation of China (325B2022, 12171318, 32500566); Shanghai Science and Technology Commission (21ZR1436300, 23XD1401900, 23DZ2290600, 24JS2810200); Science and Technology Commission of Shanghai Municipality (25ZR1402274); Medical Engineering Cross Fund of Shanghai Jiao Tong University (YG2023ZD21)
Citations: not cited yet (Europe PMC); 51 references in the paper

Abstract

Spatial transcriptomics (ST) enables mapping gene expression in native tissue context to resolve architecture and cellular interactions, but current analytical workflows rely on separate algorithms for distinct tasks. We present RGAST (Relational Graph Attention network for ST analysis), a framework that builds upon and extends our earlier HERGAST model (specifically designed for large-scale ST data analysis) for diverse downstream analysis. By introducing a relational graph attention auto-encoder, RGAST jointly models spatial proximity and gene expression similarity to capture both local and global structures in ST data. This design enables a wide range of downstream tasks within a single framework. Through comprehensive benchmarking, RGAST demonstrates superior performance in spatial domain identification across multiple platforms, improving adjusted rand index by ~10% compared to the second-best model in the dorsolateral prefrontal cortex dataset. RGAST accurately reconstructs known neuroglial interaction patterns in the mouse hypothalamus, including long-range signaling pathways that are often missed by distance-constrained methods. Moreover, RGAST also excels in boosting spatially variable gene identification accuracy, delivering more precise inference of developmental trajectories in the human cortex, and robust reconstruction of 3D tissue architectures from serial sections. Collectively, these results establish RGAST as a powerful tool for providing coherent solution to advance ST data analysis across multiple research scenarios.

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 11 matches between paragraphs and lines of code.

pypi.org/project/rgast

License: none: the authors keep all their rights
State: the link answers, verified on 27 September 2026
Evidence: the link answers
Software Heritage: not checked
Found in: “Data availability”
Not found: README, license file, CITATION.cff, environment file, tests, continuous integration, documentation
Availability: 1 check, the latest on 27 September 2026: the link answers (HTTP 200)
  • 27 September 2026: the link answers (HTTP 200)
At the source: pypi.org/project/RGAST

GYQ-form/RGAST

License: MIT
State: the link answers, verified on 27 September 2026
Evidence: files inventoried
Commit: efea2dd94f9edaf1feadbc5eb269de60ad353ca1, 17 June 2026
Languages: Python (7), Jupyter (5)
Size: 26 files, 12 scripts
Software Heritage: not checked
Found in: “Data availability”
Holds: README, license file, environment (environment.yml, setup.py), 5 notebooks
Not found: CITATION.cff, tests, continuous integration, documentation
Tools: NumPy (10 files), Scanpy (9 files), pandas (8 files), Matplotlib (7 files), scikit-learn (6 files), PyTorch Geometric (4 files), PyTorch (4 files), SciPy (4 files), anndata (3 files), seaborn (3 files), Numba (2 files), Plotly (1 file), rpy2 (1 file)
Availability: 1 check, the latest on 27 September 2026: the link answers
  • 27 September 2026: the link answers
14 files
At the source: github.com/GYQ-form/RGAST

The paper's code and data availability statement is in the Data section.

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;
  • 12 scripts, each with its path and the digest of its content;
  • 11 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

All data used in this research can be found in Supplementary Table S1. Our RGAST method is available as a Python package on PyPI at https://pypi.org/project/RGAST, free for academic use, and the source code is openly available from our GitHub repository at https://github.com/GYQ-form/RGAST.

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 1, 27 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 3 authors, 5 keywords, 7 MeSH terms, 6 funders, 49 references.

Cite

This paper

Gong, Y., Yuan, X., & Yu, Z. (2026). Empowering multifaceted analysis of spatial transcriptomics data with RGAST. Briefings in bioinformatics, 27(3), bbag298. https://doi.org/10.1093/bib/bbag298

BibTeX

@article{gong2026empowering,
author = {Gong, Yuqiao and Yuan, Xin and Yu, Zhangsheng},
title = {{Empowering multifaceted analysis of spatial transcriptomics data with RGAST}},
journal = {Briefings in bioinformatics},
year = {2026},
month = may,
volume = {27},
number = {3},
pages = {bbag298},
publisher = {Oxford University Press},
issn = {1467-5463},
doi = {10.1093/bib/bbag298},
url = {https://doi.org/10.1093/bib/bbag298},
pmid = {42302280},
pmcid = {PMC13271401}
}

RIS

TY - JOUR
AU - Gong, Yuqiao
AU - Yuan, Xin
AU - Yu, Zhangsheng
TI - Empowering multifaceted analysis of spatial transcriptomics data with RGAST
T2 - Briefings in bioinformatics
J2 - Brief Bioinform
PY - 2026
DA - 2026/05/01
VL - 27
IS - 3
SP - bbag298
SN - 1467-5463
PB - Oxford University Press
DO - 10.1093/bib/bbag298
UR - https://doi.org/10.1093/bib/bbag298
LA - en
ER -

CSL-JSON

{
"id": "10.1093/bib/bbag298",
"type": "article-journal",
"title": "Empowering multifaceted analysis of spatial transcriptomics data with RGAST",
"container-title": "Briefings in bioinformatics",
"author": [
{
"family": "Gong",
"given": "Yuqiao"
},
{
"family": "Yuan",
"given": "Xin"
},
{
"family": "Yu",
"given": "Zhangsheng"
}
],
"container-title-short": "Brief Bioinform",
"volume": "27",
"issue": "3",
"page": "bbag298",
"DOI": "10.1093/bib/bbag298",
"PMID": "42302280",
"PMCID": "PMC13271401",
"ISSN": "1467-5463",
"publisher": "Oxford University Press",
"URL": "https://doi.org/10.1093/bib/bbag298",
"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:10.1002/advs.77003 [code]
SemanticST: A Scalable Multi-Contextual Graph Learning Framework for Uncovering Spatial Niches and Robust Multi-Sample Integration in Spatial Transcriptomics.
Journal: Advanced science (Weinheim, Baden-Wurttemberg, Germany)
In common: rpy2, PyTorch Geometric, anndata, 9 other tools, methods / tools, genetics / omics, 11 references
[2] doi:10.1093/bioinformatics/btag540 [code]
Deciphering spatial heterogeneity by multimodal spatial transcriptomics modelling with SpatialModal.
Journal: Bioinformatics (Oxford, England)
In common: rpy2, PyTorch Geometric, anndata, 9 other tools, methods / tools, genetics / omics, mouse, 10 references
[3] doi:10.1038/s41592-026-03194-8 [code]
Beyond benchmarking: an expert-guided consensus approach to spatially aware clustering.
Journal: Nature methods
In common: rpy2, PyTorch Geometric, anndata, 8 other tools, methods / tools, genetics / omics, 10 references
[4] doi:10.21203/rs.3.rs-9676637/v1 [code]
A Comprehensive Benchmarking of Spatial Deconvolution and Domain Detection Methods across Diverse Tissues and Spatial Transcriptomic Technologies
Journal: Research Square (preprint)
In common: rpy2, PyTorch Geometric, anndata, 8 other tools, methods / tools, genetics / omics, 8 references
[5] doi:10.1093/bioinformatics/btag430 [code]
SPIDER: spatially integrated denoising via embedding regularization with single cell supervision.
Journal: Bioinformatics (Oxford, England)
In common: rpy2, PyTorch Geometric, anndata, 7 other tools, genetics / omics, 7 references
[6] doi:10.1371/journal.pcbi.1014346 [code]
StPedf: Cell trajectory inference of spatial transcriptomics via spatial proximity embedding and spatial density-adaptive fusion.
Journal: PLoS computational biology
In common: rpy2, anndata, Scanpy, 8 other tools, genetics / omics, 5 references
[7] doi:10.1016/j.isci.2026.117206 [code]
ReliST: A model-agnostic risk layer for spatial transcriptomics deconvolution.
Journal: iScience
In common: anndata, Scanpy, PyTorch, 6 other tools, genetics / omics, mouse, 7 references
[8] doi:10.1038/s41593-026-02293-1 [code]
Optics-free spatial genomics for mapping mammalian brain aging by IRISeq.
Journal: Nature neuroscience
In common: Numba, Scanpy, seaborn, 5 other tools, genetics / omics, mouse, 8 references
[9] doi:10.1002/advs.75969 [code]
Accurately Deciphering Tissue Heterogeneity From Spatial Multi-Modal and Multi-Omics With STransformer.
Journal: Advanced science (Weinheim, Baden-Wurttemberg, Germany)
In common: PyTorch Geometric, anndata, Scanpy, 6 other tools, genetics / omics, mouse, 6 references
[10] doi:10.1093/bib/bbag404 [code]
Navigating cell maps by deep learning integration of single-cell and spatially resolved transcriptomics.
Journal: Briefings in bioinformatics
In common: PyTorch Geometric, anndata, Scanpy, 6 other tools, genetics / omics, mouse, 6 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.

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: questions and error reports about this paper and its code, from signed-in readers and its authors. It opens with sign-in.

Reproductions: reports from readers who ran the authors' code: what they reproduced, with which environment, commit and data. It opens with sign-in.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.