Pith. sign in

REVIEW 5 major objections 5 minor 68 references

SemanticST: Spatially Informed Semantic Graph Learning for Clustering, Integration, and Scalable Analysis of Spatial Transcriptomics

T0 review · 5 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read SemanticST claims that learning multiple disentangled semantic graphs and fusing them with attention yields spatially coherent embeddings that outperform DeepST, GraphST, and IRIS by 10-20% in ARI, NMI, and trajectory fidelity, while…

desk verdict A plausible and well-benchmarked method whose headline performance claims are not yet statistically supported; needs a controlled re-run before the numbers can be trusted. read the letter →

arxiv 2506.11491 v2 pith:UV2KFQND submitted 2025-06-13 q-bio.GN cs.LG

Roxana Zahedi , Ahmadreza Argha , Nona Farbehi , Ivan Bakhshayeshi , Youqiong Ye , Nigel H. Lovell , Hamid Alinejad-Rokny This is my paper · ORCID
classification q-bio.GNcs.LG
keywords spatialtranscriptomicssemanticgraphsgraphneuralnetworksdomainidentificationbatcheffectcorrectionmin-cutclusteringXeniumscalabilitybreastcancerniches
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

SemanticST is an unsupervised graph-neural-network framework for spatial transcriptomics. It tries to establish that a single spatial dataset is better modelled by several simultaneously learned semantic graphs, each capturing a different biological relationship, than by one fixed spatial graph, and that fusing the separate embeddings with learned attention weights produces cleaner spatial domains. The paper reports consistent 10-20% gains over recent methods on clustering, trajectory, and batch-effect metrics across four platforms, and claims to be the first graph neural network that scales to 500,000-cell Xenium data via mini-batch training. If correct, it gives researchers a scalable, broadly applicable tool for spatial domain identification, batch-effect correction, and discovery of rare tissue niches.

What carries the argument

The central object is the semantic graph: a learnable weighted graph that shares the fixed spatial adjacency structure, k=5 nearest neighbours, but re-weights edges through an MLP applied to concatenated low-dimensional gene-expression features. The machinery has four parts: a cosine disentanglement loss plus a node-similarity regularizer that forces the K graphs to encode distinct aspects; K GCN encoders producing one embedding per graph; a softmax attention fusion into a single embedding; and a community-aware min-cut loss computed on soft Gumbel-Softmax community assignments that pulls neighbours into the same community and pushes communities apart. Mini-batch training over spatial graphs is what makes the framework scale to hundreds of thousands of cells, though the paper notes that the decoder output is used as the final representation on most datasets and the encoder output on Xenium.

What would settle it

Run SemanticST on DLPFC slice 151673 with K=1 and K=4: if a single semantic graph matches or beats K=4 in ARI and NMI, the multi-semantic construction is not the source of the reported gain. Separately, compare random-mini-batch training against full-batch training on a 40,000-cell dataset: if clustering fidelity drops measurably, the claim that mini-batch preserves both local and global structure is not supported.

Watch

Extended reading notes

Core claim

On the paper's own terms, SemanticST discovers that multi-semantic graph construction, building K=4 learnable weighted graphs on the same spatial five-nearest-neighbour topology with its own GCN encoder per graph, plus attention-based fusion and a community-aware min-cut loss, produces latent representations that are spatially coherent and biologically interpretable. The authors show these embeddings yield higher adjusted Rand index, normalized mutual information, and trajectory fidelity than DeepST, GraphST, SEDR, STAGATE, and IRIS across Visium, Slide-seq, Stereo-seq, and Xenium, and that the method removes batch effects in vertical and horizontal slice integration without using batch labels. They further claim that it recovers rare breast-cancer niches in Xenium data, specifically triple receptor-positive cells, distinct DCIS-to-IDC transition zones, and FOXC2-positive tumour-associated myoepithelial cells, that other methods and the original annotation miss.

Load-bearing premise

The load-bearing premise is that four semantic graphs built from the same five-nearest-neighbour spatial graph with a cosine disentanglement loss capture genuinely distinct and biologically meaningful tissue relationships, that the Moran's I heuristic used to select K on DLPFC transfers to all other datasets, and that random mini-batch sampling during training preserves enough local and global structure for the min-cut communities and cross-slice integration; if any of these fails, the claimed gains and scalability do not transfer to new datasets.

Editorial extensions

If this is right

  • Spatial domain identification can be performed on datasets of 500,000 cells or more, which current graph-neural-network baselines cannot process.
  • Batch effects between vertical or horizontal tissue sections can be removed without pre-defining batch factors, making multi-section and 3D tissue reconstruction practical.
  • Rare, clinically meaningful niches, such as triple receptor-positive tumour regions and DCIS-to-IDC transition zones, can be detected from a single Xenium section without multi-omic integration.
  • The learned attention scores per semantic graph give an interpretable decomposition of tissue regions into distinct biological programs, such as forebrain, glial, immune, and tumour programs.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A testable extension the paper leaves implicit is to treat K as a per-dataset hyperparameter selected by Moran's I instead of fixing K=4; if K=4 is not optimal for other tissues, the reported gains would vary by dataset.
  • Because each spot receives a semantic score per graph, these scores could be used as interaction features for cell-cell or niche-composition analysis; this could be checked by correlating attention weights with known ligand-receptor co-expression in the same tissue.
  • The choice to use the encoder output for Xenium and the decoder output elsewhere is a practical workaround; a principled selection rule based on gene dimensionality could make the pipeline more portable and should be tested on a mid-size gene panel.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 5 minor

Summary. SemanticST is an unsupervised graph neural network for spatial transcriptomics. It constructs K semantic graphs from a spatial k-nearest-neighbor graph, learns disentangled embeddings for each graph with GCN autoencoders, fuses them via learned attention weights, and trains with reconstruction and community-aware min-cut losses. The authors benchmark it on Visium (human DLPFC, mouse brain, breast cancer), Slide-seqV2 (mouse hippocampus), Stereo-seq (mouse olfactory bulb and embryo), and Xenium (breast cancer), reporting consistent gains in ARI, NMI, and trajectory fidelity over DeepST, GraphST, and IRIS, plus vertical and horizontal batch-effect integration. They also report biological discoveries in breast cancer Xenium data, including a rare triple receptor-positive niche and a FOXC2+ tumor-associated myoepithelial domain, and provide code and data.

Significance. The potential contribution is substantial if the empirical claims hold: SemanticST would offer an interpretable, cross-platform, scalable GNN for spatial domain identification and batch-effect correction, with public code and data. The manuscript's strengths include multi-platform benchmarking, marker-gene validation of clusters, attention-score interpretability, and the biological follow-up on the Xenium breast cancer data. However, the central claims of consistent 10-20% (or 20 percentage point) gains and first-GNN-scalable-to-500k-cells are not yet supported by statistically controlled comparisons, so the current significance is conditional on additional evidence.

major comments (5)
  1. [Abstract; Results 0.2] The abstract claims gains in 'trajectory fidelity', but no trajectory metric is defined or quantified anywhere; the only trajectory evidence is the PAGA visualization in Fig. 2F. Please add a quantitative trajectory comparison (e.g., topology or ordering score against the known laminar sequence) with values for all methods, or remove trajectory fidelity from the headline claim.
  2. [Abstract] The manuscript contains two inconsistent versions of the headline effect size: the abstract states 'consistent 20 percentage gains' while the full-text abstract states 'consistent 10-20% gains'. Since this is the central quantitative claim, please harmonize the wording and provide per-dataset, per-metric numbers in a summary table.
  3. [Method 0.10-0.11; Results 0.5] Reported superiority is based on single runs without seed-level repeats or confidence intervals, while several key hyperparameters were selected on the benchmark data themselves: k=5 was chosen as the value with 'best performance' (Section 0.10), K=4 was chosen using Moran's I on DLPFC (Section 0.11.1), and cluster counts or clustering resolutions differ across datasets (22 clusters for the embryo, Louvain resolution 1 in Fig. 3B vs 0.3 in Fig. 3D). Please provide repeated-run statistics with significance tests and sensitivity analyses for k, K, beta1, and beta2; this is necessary to support the word 'consistent'.
  4. [Method 0.11.1-0.11.2; Eq. (13)] The method definition has several notation and dimension issues that block independent implementation: in Eq. (3) the weight matrix W_s is multiplied before the feature matrix Z_s, which is not the standard GCN ordering; Eq. (4) defines d_k as a graph-level descriptor vector but Eq. (6) uses per-node descriptors d_i^k and Eq. (8) multiplies d_k as if it were a scalar edge weight; and Eq. (13) uses g_m in the denominator without ever defining it. Please specify all tensor dimensions and operation order, and define g_m.
  5. [Results 0.7; Method 0.17] The scalability claim is not fully supported by the presented comparisons. The 500,000-cell Xenium evaluation includes only IRIS and Banksy, both non-deep-learning methods, and does not compare against SGCAST, the mini-batch GNN cited in the introduction, or any other scalable GNN baseline. In addition, Section 0.17 asserts that mini-batch training preserves local and global structure but provides no experiment comparing mini-batch against full-batch performance or varying batch size. Please add such a comparison or temper the 'first GNN scalable to Xenium' claim.
minor comments (5)
  1. [Results 0.2] In the caption of Fig. 2D, 'HMO' should be 'HOM'; also the sentence 'Despite constructing four semantic graphs by aggregating information in the embedding space (Fig. 2E)' appears to cite the runtime panel, not the semantic-graph information.
  2. [Results 0.7.4] The text refers to Fig. 4F and Fig. 4G/H when describing the Mann-Whitney U test results and the FOXC2/CDH1 spatial plots; these should be Fig. 5F and Fig. 5G/H.
  3. [Method 0.11.3] 'fussed latent representations' should be 'fused', and 'an additional layer of interoperability' should likely be 'interpretability'.
  4. [Method 0.9] The preprocessing section states that gene/cell filtering is 'dependent on the specific ST technology' and then specifies exceptions for Visium and Xenium, but the exact filtering applied to each benchmark platform is not listed; a per-dataset preprocessing table would remove ambiguity.
  5. [Various] There are multiple typographical issues, including 'T GF β' for TGF-β, 'F ASN' for FASN, and 'Nspot/cell' spacing; a careful proofread is recommended.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation found: SemanticST's semantic graphs and embeddings are learned unsupervised and benchmarks are external; the K=4 Moran's I choice is hyperparameter tuning, not a fitted prediction.

full rationale

SemanticST's derivation chain is self-contained with respect to the quantities it predicts. The semantic graphs are produced by learned MLP edge weights over the shared spatial kNN graph (Eq. 2), embeddings are learned by GCN encoders with reconstruction and min-cut losses (Eqs. 8, 12, 13), and the fused representation is an attention-weighted sum (Eq. 9); none of these equations is defined in terms of the benchmark labels or of the ARI/NMI values reported. The clustering benchmarks compare SemanticST embeddings against external methods (DeepST, GraphST, STAGATE, SEDR, SpaGCN, IRIS, Banksy), so the headline gains are empirical claims about independent data, not identities. The only model-selection feedback that touches a benchmark is the choice of K=4 via mean Moran's I 'validated on the DLPFC dataset' (Section 0.11.1) and then applied to DLPFC reporting; however, Moran's I is an unsupervised spatial-autocorrelation statistic computed from gene expression and graph weights, not a fit to layer labels, so the DLPFC ARI is not forced by construction. The abstract's unquantified 'trajectory fidelity' claim and the absence of seed-level repeats are evaluation and robustness limitations, not circular derivations, and no self-citation chain is load-bearing. No circular step could be identified with the required specific reduction.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The central claim rests on several tuned constants and domain assumptions about what spatial graphs represent. No new physical entities are introduced. The free parameters are mostly hyperparameters chosen on the benchmark data rather than derived from first principles.

free parameters (6)
  • k (nearest-neighbor count) = 5
    Set in Section 0.10 because 'this value provided the best performance'; affects all graph constructions and hence all benchmarks.
  • K (number of semantic graphs) = 4
    Selected in Section 0.11.1 using mean Moran's I and losssemantic on the DLPFC dataset, then applied to all other datasets.
  • lambda (Lnode regularization weight) = not reported
    Appears in Eq 7 but no value is given, so the semantic graph loss is not fully specified.
  • beta1 (reconstruction loss weight) = 10
    Set in Section 0.12.1; controls balance between reconstruction and min-cut objectives.
  • beta2 (min-cut loss weight) = 0.1
    Set in Section 0.12.1; controls community-separation pressure.
  • M (community count) = hidden dimension
    Section 0.12.1 sets the number of communities equal to the hidden dimension rather than an inferred cluster count.
assumptions (4)
  • domain assumption Spatial kNN graph with k=5 captures biologically meaningful tissue relationships across Visium, Slide-seq, Stereo-seq, and Xenium.
    Used to construct all graphs; no data-driven validation across platforms; k is tuned not derived.
  • domain assumption A fixed set of K semantic graphs learned by an MLP over the same adjacency matrix encodes distinct, disentangled biological relation spaces.
    Section 0.11.1; the disentanglement is enforced only by cosine loss Ldis and homogeneity loss Lnode.
  • domain assumption Decoder output He is a valid clustered representation except for Xenium, where encoder output replaces it.
    Section 0.13 and Discussion; the decoder choice is post hoc per dataset.
  • domain assumption Block-diagonal concatenation of graph adjacency matrices removes batch effects without batch labels.
    Section 0.16; the method is asserted to implicitly correct batch effects.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SemanticST: Spatially Informed Semantic Graph Learning for Clustering, Integration, and Scalable Analysis of Spatial Transcriptomics." pith.science (2026). https://pith.science/paper/UV2KFQND

@misc{pith2026250611491,
  author       = {Pith},
  title        = {Pith review of: SemanticST: Spatially Informed Semantic Graph Learning for Clustering, Integration, and Scalable Analysis of Spatial Transcriptomics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UV2KFQND}},
  note         = {Machine review of arXiv:2506.11491}
}
read the original abstract

Spatial transcriptomics (ST) technologies enable gene expression profiling with spatial resolution, offering unprecedented insights into tissue organization and disease heterogeneity. However, current analysis methods often struggle with noisy data, limited scalability, and inadequate modelling of complex cellular relationships. We present SemanticST, a biologically informed, graph-based deep learning framework that models diverse cellular contexts through multi-semantic graph construction. SemanticST builds multiple context-specific graphs capturing spatial proximity, gene expression similarity, and tissue domain structure, and learns disentangled embeddings for each. These are fused using an attention-inspired strategy to yield a unified, biologically meaningful representation. A community-aware min-cut loss improves robustness over contrastive learning, particularly in sparse ST data. SemanticST supports mini-batch training, making it the first graph neural network scalable to large-scale datasets such as Xenium (500,000 cells). Benchmarking across four platforms (Visium, Slide-seq, Stereo-seq, Xenium) and multiple human and mouse tissues shows consistent 20 percentage gains in ARI, NMI, and trajectory fidelity over DeepST, GraphST, and IRIS. In re-analysis of breast cancer Xenium data, SemanticST revealed rare and clinically significant niches, including triple receptor-positive clusters, spatially distinct DCIS-to-IDC transition zones, and FOXC2 tumour-associated myoepithelial cells, suggesting non-canonical EMT programs with stem-like features. SemanticST thus provides a scalable, interpretable, and biologically grounded framework for spatial transcriptomics analysis, enabling robust discovery across tissue types and diseases, and paving the way for spatially resolved tissue atlases and next-generation precision medicine.

Figures

Figures reproduced from arXiv: 2506.11491 by the authors.

Figure 4
Figure 4. Application of SemanticST on Visium ST data of human breast cancer. [PITH_FULL_IMAGE:figures/full_fig_p013_4.png] view at source ↗
Figure 5
Figure 5. Xenium ST analysis of Breast Cancer. A. Spatial domain identification using data from Janesick et al. (annotation), Banksy, IRIS, and SemanticST. B. Heatmap displaying well-known genes across all spatial domains identified by SemanticST. C. Dot plot illustrating the expression of cancer-related genes across all spatial domains detected by SemanticST. D. Tissue classification based on SemanticST-detected domains of m… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

68 extracted references · 63 canonical work pages

  1. [1]

    An introduction to spatial transcriptomics for biomedical research

    Cameron G Williams, Hyun Jae Lee, Takahiro Asatsuma, Roser Vento-Tormo, and Ashraful Haque. An introduction to spatial transcriptomics for biomedical research. Genome Medicine, 14(1):68, 2022

    work page 2022

  2. [2]

    Holt and Simon L

    Christine E. Holt and Simon L. Bullock. Subcellular mrna localization in animal cells and why it matters. Science, 326(5957):1212–1216, 2009

    work page 2009

  3. [3]

    Single-cell in situ rna profiling by sequential hybridization

    Eric Lubeck, Ahmet F Coskun, Timur Zhiyentayev, Mubhij Ahmad, and Long Cai. Single-cell in situ rna profiling by sequential hybridization. Nature methods, 11(4):360–361, 2014

    work page 2014

  4. [4]

    Transcriptome-scale spatial gene expression in the human dorsolateral prefrontal cortex

    Kristen R Maynard, Leonardo Collado-Torres, Lukas M Weber, Cedric Uytingco, Brianna K Barry, Stephen R Williams, Joseph L Catallini, Matthew N Tran, Zachary Besich, Madhavi Tippani, et al. Transcriptome-scale spatial gene expression in the human dorsolateral prefrontal cortex. Nature neuro- science, 24(3):425–436, 2021

    work page 2021

  5. [5]

    Rnaimaging

    BoettigerAN ChenKH et al. Rnaimaging. spatiallyresolved, highlymultiplexedrnaprofilinginsinglecells. Science, 348:6233, 2015

    work page 2015

  6. [6]

    Three-dimensional intact-tissue se- quencing of single-cell transcriptional states

    Xiao Wang, William E Allen, Matthew A Wright, Emily L Sylwestrak, Nikolay Samusik, Sam Vesuna, Kathryn Evans, Cindy Liu, Charu Ramakrishnan, Jia Liu, et al. Three-dimensional intact-tissue se- quencing of single-cell transcriptional states. Science, 361(6400):eaat5691, 2018

    work page 2018

  7. [7]

    Transcriptome-scale super-resolved imaging in tissues by rna seqfish+

    Chee-Huat Linus Eng, Michael Lawson, Qian Zhu, Ruben Dries, Noushin Koulena, Yodai Takei, Jina Yun, Christopher Cronin, Christoph Karp, Guo-Cheng Yuan, et al. Transcriptome-scale super-resolved imaging in tissues by rna seqfish+. Nature, 568(7751):235–239, 2019

    work page 2019

  8. [8]

    Visualization and analysis of gene expression in tissue sections by spatial transcriptomics

    Patrik L St ˚ ahl, Fredrik Salm´ en, Sanja Vickovic, Anna Lundmark, Jos´ e Fern´ andez Navarro, Jens Mag- nusson, Stefania Giacomello, Michaela Asp, Jakub O Westholm, Mikael Huss, et al. Visualization and analysis of gene expression in tissue sections by spatial transcriptomics. Science, 353(6294):78–82, 2016

    work page 2016

Show all 68 references
  1. [9]

    Highly sensitive spatial transcriptomics at near-cellular resolution with slide-seqv2

    Robert R Stickels, Evan Murray, Pawan Kumar, Jilong Li, Jamie L Marshall, Daniela J Di Bella, Paola Arlotta, Evan Z Macosko, and Fei Chen. Highly sensitive spatial transcriptomics at near-cellular resolution with slide-seqv2. Nature biotechnology, 39(3):313–319, 2021

    2021

  2. [10]

    High- definition spatial transcriptomics for in situ tissue profiling

    Sanja Vickovic, G¨ okcen Eraslan, Fredrik Salm´ en, Johanna Klughammer, Linnea Stenbeck, Denis Schapiro, Tarmo ¨Aij¨ o, Richard Bonneau, Ludvig Bergenstr ˚ ahle, Jos´ e Fernand´ ez Navarro, et al. High- definition spatial transcriptomics for in situ tissue profiling. Nature me...

    2019

  3. [11]

    Laser-capture microdissection

    Virginia Espina, Julia D Wulfkuhle, Valerie S Calvert, Amy VanMeter, Weidong Zhou, George Coukos, David H Geho, Emanuel F Petricoin III, and Lance A Liotta. Laser-capture microdissection. Nature protocols, 1(2):586–603, 2006

    2006

  4. [12]

    Spatial recon- struction of single-cell gene expression data

    Rahul Satija, Jeffrey A Farrell, David Gennert, Alexander F Schier, and Aviv Regev. Spatial recon- struction of single-cell gene expression data. Nature biotechnology, 33(5):495–502, 2015

    2015

  5. [13]

    Giotto: a toolbox for integrative analysis and visualization of spatial expression data

    Ruben Dries, Qian Zhu, Rui Dong, Chee-Huat Linus Eng, Huipeng Li, Kan Liu, Yuntian Fu, Tianxiao Zhao, Arpan Sarkar, Feng Bao, et al. Giotto: a toolbox for integrative analysis and visualization of spatial expression data. Genome biology, 22:1–31, 2021

    2021

  6. [14]

    Spagcn: Integrating gene expression, spatial location and histology to identify spatial domains and spatially variable genes by graph convolutional network

    Jian Hu, Xiangjie Li, Kyle Coleman, Amelia Schroeder, Nan Ma, David J Irwin, Edward B Lee, Russell T Shinohara, and Mingyao Li. Spagcn: Integrating gene expression, spatial location and histology to identify spatial domains and spatially variable genes by graph convolutional n...

    2021

  7. [15]

    Deepst: identifying spatial domains in spatial transcriptomics by deep learning

    Chang Xu, Xiyun Jin, Songren Wei, Pingping Wang, Meng Luo, Zhaochun Xu, Wenyi Yang, Yideng Cai, Lixing Xiao, Xiaoyu Lin, et al. Deepst: identifying spatial domains in spatial transcriptomics by deep learning. Nucleic Acids Research, 50(22):e131–e131, 2022

    2022

  8. [16]

    Deciphering spatial domains from spatially resolved transcriptomics with an adaptive graph attention auto-encoder

    Kangning Dong and Shihua Zhang. Deciphering spatial domains from spatially resolved transcriptomics with an adaptive graph attention auto-encoder. Nature communications, 13(1):1739, 2022

    2022

  9. [17]

    Unsupervised spatially embedded deep representation of spatial transcriptomics

    Hang Xu, Huazhu Fu, Yahui Long, Kok Siong Ang, Raman Sethi, Kelvin Chong, Mengwei Li, Rom Ud- damvathanak, Hong Kai Lee, Jingjing Ling, et al. Unsupervised spatially embedded deep representation of spatial transcriptomics. Genome Medicine, 16(1):12, 2024

    2024

  10. [18]

    Robust mapping of spatiotemporal trajectories and cell–cell interactions in healthy and diseased tissues

    Duy Pham, Xiao Tan, Brad Balderson, Jun Xu, Laura F Grice, Sohye Yoon, Emily F Willis, Minh Tran, Pui Yeng Lam, Arti Raghubar, et al. Robust mapping of spatiotemporal trajectories and cell–cell interactions in healthy and diseased tissues. Nature communications, 14(1):7739, 2023

    2023

  11. [19]

    Spatially informed clustering, integra- tion, and deconvolution of spatial transcriptomics with graphst

    Yahui Long, Kok Siong Ang, Mengwei Li, Kian Long Kelvin Chong, Raman Sethi, Chengwei Zhong, Hang Xu, Zhiwei Ong, Karishma Sachaphibulkij, Ao Chen, et al. Spatially informed clustering, integra- tion, and deconvolution of spatial transcriptomics with graphst. Nature Communicati...

    2023

  12. [20]

    const: an interpretable multi-modal contrastive learning framework for spatial transcriptomics

    Yongshuo Zong, Tingyang Yu, Xuesong Wang, Yixuan Wang, Zhihang Hu, and Yu Li. const: an interpretable multi-modal contrastive learning framework for spatial transcriptomics. BioRxiv, pages 2022–01, 2022

    2022

  13. [21]

    Deep graph infomax

    Petar Velickovic, William Fedus, William L Hamilton, Pietro Li` o, Yoshua Bengio, and R Devon Hjelm. Deep graph infomax. ICLR (Poster) , 2(3):4, 2019

    2019

  14. [22]

    Sgcast: symmetric graph convolutional auto-encoder for scalable and accurate study of spatial transcriptomics

    Jinzhao Li, Jiong Wang, and Zhixiang Lin. Sgcast: symmetric graph convolutional auto-encoder for scalable and accurate study of spatial transcriptomics. Briefings in Bioinformatics , 25(1):bbad490, 2024

    2024

  15. [23]

    nearest neighbor

    Kevin Beyer, Jonathan Goldstein, Raghu Ramakrishnan, and Uri Shaft. When is “nearest neighbor” meaningful? In Database Theory—ICDT’99: 7th International Conference Jerusalem, Israel, January 10–12, 1999 Proceedings 7, pages 217–235. Springer, 1999

    1999

  16. [24]

    Museum of spatial transcriptomics

    Lambda Moses and Lior Pachter. Museum of spatial transcriptomics. Nature methods, 19(5):534–546, 2022

    2022

  17. [25]

    Computational methods for alignment and integration of spatially resolved transcriptomics data

    Yuyao Liu and Can Yang. Computational methods for alignment and integration of spatially resolved transcriptomics data. Computational and Structural Biotechnology Journal , 2024

    2024

  18. [26]

    Computational principles and challenges in single-cell data integration

    Ricard Argelaguet, Anna SE Cuomo, Oliver Stegle, and John C Marioni. Computational principles and challenges in single-cell data integration. Nature biotechnology, 39(10):1202–1215, 2021. 27

    2021

  19. [27]

    Fast, sensitive and accurate integration of single-cell data with harmony

    Ilya Korsunsky, Nghia Millard, Jean Fan, Kamil Slowikowski, Fan Zhang, Kevin Wei, Yuriy Baglaenko, Michael Brenner, Po-ru Loh, and Soumya Raychaudhuri. Fast, sensitive and accurate integration of single-cell data with harmony. Nature methods, 16(12):1289–1296, 2019

    2019

  20. [28]

    Deep generative modeling for single-cell transcriptomics

    Romain Lopez, Jeffrey Regier, Michael B Cole, Michael I Jordan, and Nir Yosef. Deep generative modeling for single-cell transcriptomics. Nature methods, 15(12):1053–1058, 2018

    2018

  21. [29]

    Deep mincut: Learning node embeddings by detecting communities

    Chi Thang Duong, Thanh Tam Nguyen, Trung-Dung Hoang, Hongzhi Yin, Matthias Weidlich, and Quoc Viet Hung Nguyen. Deep mincut: Learning node embeddings by detecting communities. Pattern Recognition, 134:109126, 2023

    2023

  22. [30]

    Review of brain functioning in depression for semantic processing and verbal fluency

    Heide Klumpp and Patricia Deldin. Review of brain functioning in depression for semantic processing and verbal fluency. International Journal of Psychophysiology , 75(2):77–85, 2010

    2010

  23. [31]

    Large-scale exome sequencing study implicates both developmental and functional changes in the neurobiology of autism

    F Kyle Satterstrom, Jack A Kosmicki, Jiebiao Wang, Michael S Breen, Silvia De Rubeis, Joon-Yong An, Minshi Peng, Ryan Collins, Jakob Grove, Lambertus Klei, et al. Large-scale exome sequencing study implicates both developmental and functional changes in the neurobiology of aut...

    2020

  24. [32]

    Common schizophrenia alleles are enriched in mutation-intolerant genes and in regions under strong background selection

    Antonio F Pardi˜ nas, Peter Holmans, Andrew J Pocklington, Valentina Escott-Price, Stephan Ripke, Noa Carrera, Sophie E Legge, Sophie Bishop, Darren Cameron, Marian L Hamshere, et al. Common schizophrenia alleles are enriched in mutation-intolerant genes and in regions under s...

    2018

  25. [33]

    Genome-wide association study identifies 30 loci associated with bipolar disorder

    Eli A Stahl, Gerome Breen, Andreas J Forstner, Andrew McQuillin, Stephan Ripke, Vassily Trubetskoy, Manuel Mattheisen, Yunpeng Wang, Jonathan RI Coleman, H´ el´ ena A Gaspar, et al. Genome-wide association study identifies 30 loci associated with bipolar disorder. Nature genet...

    2019

  26. [34]

    Fast unfold- ing of communities in large networks

    Vincent D Blondel, Jean-Loup Guillaume, Renaud Lambiotte, and Etienne Lefebvre. Fast unfold- ing of communities in large networks. Journal of statistical mechanics: theory and experiment , 2008(10):P10008, 2008

    2008

  27. [35]

    Neurogranin enhances synaptic strength through its interaction with calmodulin

    Ling Zhong, Tiffani Cherry, Christine E Bies, Matthew A Florence, and Nashaat Z Gerges. Neurogranin enhances synaptic strength through its interaction with calmodulin. The EMBO journal , 28(19):3027– 3039, 2009

    2009

  28. [36]

    Association between nrgn gene polymorphism and resting-state hippocampal functional connectivity in schizophrenia

    Yifan Zhang, Xiaohong Gong, Zhiyang Yin, Lingling Cui, Jian Yang, Pengshuo Wang, Yifang Zhou, Xiaowei Jiang, Shengnan Wei, Fei Wang, et al. Association between nrgn gene polymorphism and resting-state hippocampal functional connectivity in schizophrenia. BMC psychiatry , 19:1–9, 2019

    2019

  29. [37]

    Expression of c1ql3 in discrete neuronal populations controls efferent synapse numbers and diverse behaviors

    David C Martinelli, Kylie S Chew, Astrid Rohlmann, Matthew Y Lum, Susanne Ressl, Samer Hattar, Axel T Brunger, Markus Missler, and Thomas C S¨ udhof. Expression of c1ql3 in discrete neuronal populations controls efferent synapse numbers and diverse behaviors. Neuron, 91(5):103...

    2016

  30. [38]

    Neurod6 expression defines new retinal amacrine cell subtypes and regulates their fate

    Jeremy N Kay, P Emanuela Voinescu, Monica W Chu, and Joshua R Sanes. Neurod6 expression defines new retinal amacrine cell subtypes and regulates their fate. Nature neuroscience, 14(8):965–972, 2011

    2011

  31. [39]

    An rna-sequencing transcriptome and splicing database of glia, neurons, and vascular cells of the cerebral cortex

    Ye Zhang, Kenian Chen, Steven A Sloan, Mariko L Bennett, Anja R Scholze, Sean O’Keeffe, Hemali P Phatnani, Paolo Guarnieri, Christine Caneda, Nadine Ruderisch, et al. An rna-sequencing transcriptome and splicing database of glia, neurons, and vascular cells of the cerebral cor...

    2014

  32. [40]

    Purification and characteriza- tion of progenitor and mature human astrocytes reveals transcriptional and functional differences with mouse

    Ye Zhang, Steven A Sloan, Laura E Clarke, Christine Caneda, Colton A Plaza, Paul D Blumenthal, Hannes Vogel, Gary K Steinberg, Michael SB Edwards, Gordon Li, et al. Purification and characteriza- tion of progenitor and mature human astrocytes reveals transcriptional and functi...

    2016

  33. [41]

    Spatiotemporal transcriptomic atlas of mouse organogenesis using dna nanoball-patterned arrays

    Ao Chen, Sha Liao, Mengnan Cheng, Kailong Ma, Liang Wu, Yiwei Lai, Xiaojie Qiu, Jin Yang, Jiang- shan Xu, Shijie Hao, et al. Spatiotemporal transcriptomic atlas of mouse organogenesis using dna nanoball-patterned arrays. Cell, 185(10):1777–1792, 2022

    2022

  34. [42]

    Gene expression profiling of atp-binding cassette (abc) transporters as a predictor of the pathologic response to neoadjuvant chemotherapy in breast cancer patients

    Sarah Park, Chikako Shimizu, Tatsu Shimoyama, Masayuki Takeda, Masashi Ando, Tsutomu Kohno, Noriyuki Katsumata, Yoon-Koo Kang, Kazuto Nishio, and Yasuhiro Fujiwara. Gene expression profiling of atp-binding cassette (abc) transporters as a predictor of the pathologic response t...

    2006

  35. [43]

    Spatial transcriptomics reveals distinct and conserved tumor core and edge architectures that predict survival and targeted therapy response

    Rohit Arora, Christian Cao, Mehul Kumar, Sarthak Sinha, Ayan Chanda, Reid McNeil, Divya Samuel, Rahul K Arora, T Wayne Matthews, Shamir Chandarana, et al. Spatial transcriptomics reveals distinct and conserved tumor core and edge architectures that predict survival and targete...

    2023

  36. [44]

    The emerging role of xbp1 in cancer

    Shanshan Chen, Jing Chen, Xin Hua, Yue Sun, Rui Cui, Jun Sha, and Xiaoli Zhu. The emerging role of xbp1 in cancer. Biomedicine & Pharmacotherapy, 127:110069, 2020

    2020

  37. [45]

    Prognostic impact of agr3 protein expression in breast cancer: A systematic review and meta-analysis

    Carolina Le˜ ao de Moraes, Carolina Rodrigues Mendon¸ ca, Nat´ alia Cruz e Melo, Fernanda Sardinha de Abreu Tacon, Jair Pereira de Melo Junior, and Waldemar Naves do Amaral. Prognostic impact of agr3 protein expression in breast cancer: A systematic review and meta-analysis. R...

    2023

  38. [46]

    The biomarker he4 (wfdc2) promotes a pro-angiogenic and immunosuppressive tumor microenvironment via regulation of stat3 target genes

    Nicole E James, Jenna B Emerson, Ashley D Borgstadt, Lindsey Beffa, Matthew T Oliver, Virginia Hovanesian, Anze Urh, Rakesh K Singh, Rachael Rowswell-Turner, Paul A DiSilvestro, et al. The biomarker he4 (wfdc2) promotes a pro-angiogenic and immunosuppressive tumor microenviron...

    2020

  39. [47]

    Molecular portraits of human breast tumours

    Charles M Perou, Therese Sørlie, Michael B Eisen, Matt Van De Rijn, Stefanie S Jeffrey, Christian A Rees, Jonathan R Pollack, Douglas T Ross, Hilde Johnsen, Lars A Akslen, et al. Molecular portraits of human breast tumours. nature, 406(6797):747–752, 2000

    2000

  40. [48]

    Tissue-based map of the human proteome

    Mathias Uhl´ en, Linn Fagerberg, Bj¨ orn M Hallstr¨ om, Cecilia Lindskog, Per Oksvold, Adil Mardinoglu, ˚Asa Sivertsson, Caroline Kampf, Evelina Sj¨ ostedt, Anna Asplund, et al. Tissue-based map of the human proteome. Science, 347(6220):1260419, 2015

    2015

  41. [49]

    Aebp1 con- tributes to breast cancer progression by facilitating cell proliferation, migration, invasion, and blocking apoptosis

    Jin Li, Yanyun Ruan, Chenhui Zheng, Yue Pan, Bangyi Lin, Qi Chen, and Zhibao Zheng. Aebp1 con- tributes to breast cancer progression by facilitating cell proliferation, migration, invasion, and blocking apoptosis. Discovery Medicine, 35(174):45–56, 2023

    2023

  42. [50]

    High resolution mapping of the tumor microenvironment using integrated single-cell, spatial and in situ analysis

    Amanda Janesick, Robert Shelansky, Andrew D Gottscho, Florian Wagner, Stephen R Williams, Mor- gane Rouault, Ghezal Beliakoff, Carolyn A Morrison, Michelli F Oliveira, Jordan T Sicherman, et al. High resolution mapping of the tumor microenvironment using integrated single-cell...

    2023

  43. [51]

    Accurate and efficient integrative reference-informed spatial domain detection for spatial transcriptomics

    Ying Ma and Xiang Zhou. Accurate and efficient integrative reference-informed spatial domain detection for spatial transcriptomics. Nature Methods, 21(7):1231–1244, 2024

    2024

  44. [52]

    Banksy unifies cell typing and tissue domain segmen- tation for scalable spatial omics data analysis

    Vipul Singhal, Nigel Chou, Joseph Lee, Yifei Yue, Jinyue Liu, Wan Kee Chock, Li Lin, Yun-Ching Chang, Erica Mei Ling Teo, Jonathan Aow, et al. Banksy unifies cell typing and tissue domain segmen- tation for scalable spatial omics data analysis. Nature genetics, 56(3):431–441, 2024

    2024

  45. [53]

    Ductal carcinoma in situ: molecular changes accompanying disease progression

    Gemma M Wilson, Phuong Dinh, Nirmala Pathmanathan, and J Dinny Graham. Ductal carcinoma in situ: molecular changes accompanying disease progression. Journal of mammary gland biology and neoplasia, 27(1):101–131, 2022

    2022

  46. [54]

    Ceacam5 and ceacam6 are major target genes for smad3-mediated tgf-β signaling

    Sang Uk Han, Tae Hee Kwak, Kyu Hee Her, Young-Hun Cho, Chul-won Choi, Ho Jae Lee, Suntaek Hong, Yoon Soo Park, YS Kim, Tae Aug Kim, et al. Ceacam5 and ceacam6 are major target genes for smad3-mediated tgf-β signaling. Oncogene, 27(5):675–683, 2008. 29

    2008

  47. [55]

    Tumor-associated myoepithelial cells promote the invasive progression of ductal carci- noma in situ through activation of tgfβ signaling

    Pang-Kuo Lo, Yongshu Zhang, Yuan Yao, Benjamin Wolfson, Justine Yu, Shu-Yan Han, Nadire Duru, and Qun Zhou. Tumor-associated myoepithelial cells promote the invasive progression of ductal carci- noma in situ through activation of tgfβ signaling. Journal of Biological Chemistry...

    2017

  48. [56]

    Mann-whitney u test

    Patrick E McKnight and Julius Najab. Mann-whitney u test. The Corsini encyclopedia of psychology , pages 1–1, 2010

    2010

  49. [57]

    Foxc2 expres- sion links epithelial–mesenchymal transition and stem cell properties in breast cancer

    Brett G Hollier, Agata A Tinnirello, Steven J Werden, Kurt W Evans, Joseph H Taube, Tapasree Roy Sarkar, Nathalie Sphyris, Maryam Shariati, Sreedevi V Kumar, Venkata L Battula, et al. Foxc2 expres- sion links epithelial–mesenchymal transition and stem cell properties in breast...

    1981

  50. [58]

    Allen brain atlas, 2006

    Developing Mouse Brain Atlas. Allen brain atlas, 2006

    2006

  51. [59]

    Defining cell types and states with single-cell genomics

    Cole Trapnell. Defining cell types and states with single-cell genomics. Genome research, 25(10):1491– 1498, 2015

    2015

  52. [60]

    Cancer statistics, 2024

    Rebecca L Siegel, Angela N Giaquinto, and Ahmedin Jemal. Cancer statistics, 2024. CA: a cancer journal for clinicians , 74(1):12–49, 2024

    2024

  53. [61]

    Scanpy: large-scale single-cell gene expression data analysis

    F Alexander Wolf, Philipp Angerer, and Fabian J Theis. Scanpy: large-scale single-cell gene expression data analysis. Genome biology, 19:1–5, 2018

    2018

  54. [62]

    Graphmixup: Improving class-imbalanced node classification on graphs by self-supervised context prediction

    Lirong Wu, Haitao Lin, Zhangyang Gao, Cheng Tan, Stan Li, et al. Graphmixup: Improving class-imbalanced node classification on graphs by self-supervised context prediction. arXiv preprint arXiv:2106.11133, 2021

    2021 arXiv

  55. [63]

    Neural machine translation with gumbel-greedy decoding

    Jiatao Gu, Daniel Jiwoong Im, and Victor OK Li. Neural machine translation with gumbel-greedy decoding. In Proceedings of the AAAI conference on artificial intelligence , volume 32, 2018

    2018

  56. [64]

    Adam: A method for stochastic optimization

    P Kingma Diederik. Adam: A method for stochastic optimization. In International Conference on Learning Representations, 2015

    2015

  57. [65]

    mclust version 4 for r: normal mixture modeling for model-based clustering, classification, and density estimation

    Chris Fraley, Adrian E Raftery, T Brendan Murphy, and Luca Scrucca. mclust version 4 for r: normal mixture modeling for model-based clustering, classification, and density estimation. Technical report, Citeseer, 2012

    2012

  58. [66]

    Paga: graph abstraction reconciles clustering with trajectory inference through a topology preserving map of single cells

    F Alexander Wolf, Fiona K Hamey, Mireya Plass, Jordi Solana, Joakim S Dahlin, Berthold G¨ ottgens, Nikolaus Rajewsky, Lukas Simon, and Fabian J Theis. Paga: graph abstraction reconciles clustering with trajectory inference through a topology preserving map of single cells. Gen...

    2019

  59. [67]

    Alignment and integration of spatial transcriptomics data

    Ron Zeira, Max Land, Alexander Strzalkowski, and Benjamin J Raphael. Alignment and integration of spatial transcriptomics data. Nature Methods, 19(5):567–575, 2022

    2022

  60. [68]

    Inductive representation learning on large graphs

    Will Hamilton, Zhitao Ying, and Jure Leskovec. Inductive representation learning on large graphs. Advances in neural information processing systems , 30, 2017. 30

    2017

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.