arxiv: 2605.12662 · v1 · submitted 2026-05-12 · 💻 cs.LG · q-bio.GN

Recognition: 2 theorem links

· Lean Theorem

scShapeBench: Discovering geometry from high dimensional scRNAseq data

Andrew J Steindl , Jo\~ao Felipe Rocha , Brian Tshilengi Di Bassinga , Zachary Warren , Matthew Scicluna , C\'esar Miguel Valdez C\'ordova , Shabarni Gupta , Leire Torices

show 8 more authors

Daniel Neumann Timothy J. Mann Ihuan Gunawan Dhananjay Bhaskar John G Lock Christine L Chaffer Guy Wolf Smita Krishnaswamy

Authors on Pith no claims yet

Pith reviewed 2026-05-14 21:17 UTC · model grok-4.3

classification 💻 cs.LG q-bio.GN

keywords datadatasetsanalysisshapesingle-celldetectionpipelinescreebtower

0 comments

The pith

scReebTower extracts Reeb graphs from diffusion geometry to classify single-cell data shapes more accurately than PAGA or Mapper.

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

The paper creates scShapeBench, a collection of synthetic point clouds sampled from known skeleton graphs plus real single-cell datasets labeled by experts into four shapes: clusters, single trajectories, multi-branching, and archetypal. It then presents scReebTower, which computes Reeb graphs on diffusion distances to recover these shapes and thereby select appropriate downstream pipelines. This setup replaces manual visual inspection with an automated step that matches data geometry to tools such as Louvain clustering or trajectory inference. If the method works as described, analysis of new single-cell experiments becomes more consistent because the chosen pipeline matches the actual topology rather than an assumed one. Results on both synthetic and real data show higher accuracy than existing baselines under topology-aware metrics.

Core claim

scShapeBench supplies ground-truth synthetic data and expert-annotated real single-cell datasets grouped into four discrete geometric categories, while scReebTower builds Reeb graphs from diffusion geometry to recover those categories and link visualization directly to pipeline selection, outperforming PAGA and Mapper on the provided evaluation metrics.

What carries the argument

scReebTower, which constructs Reeb graphs on diffusion distances to represent the shape of high-dimensional single-cell point clouds and connect that representation to downstream analysis choice.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

If shape detection becomes reliable, the same benchmark format could be reused to test automated pipeline selection in other high-dimensional domains such as spatial transcriptomics or flow cytometry.
Accurate shape labels would allow direct measurement of how often current analysis choices mismatch data geometry and therefore quantify hidden bias in published single-cell studies.
The four-category taxonomy could be expanded by measuring continuous variation between shapes rather than forcing discrete labels on borderline datasets.

Load-bearing premise

Expert annotations of real single-cell datasets into the four discrete shape categories are accurate, consistent, and sufficient to represent the geometries that matter for downstream analysis.

What would settle it

A fresh collection of single-cell datasets in which independent experts produce conflicting shape labels or in which scReebTower no longer records higher topology-aware scores than PAGA and Mapper.

Figures

Figures reproduced from arXiv: 2605.12662 by Andrew J Steindl, Brian Tshilengi Di Bassinga, C\'esar Miguel Valdez C\'ordova, Christine L Chaffer, Daniel Neumann, Dhananjay Bhaskar, Guy Wolf, Ihuan Gunawan, Jo\~ao Felipe Rocha, John G Lock, Leire Torices, Matthew Scicluna, Shabarni Gupta, Smita Krishnaswamy, Timothy J. Mann, Zachary Warren.

**Figure 2.** Figure 2: Single-cell shape recovery framework. The goal of shape recovery is to infer a graph representation of the underlying organization of high-dimensional data. (a) A latent graph G describes the underlying organization of the data. (b) Samples from this structure produce the observed noisy high-dimensional point cloud X. (c) A shape recovery method constructs a graph S from the observed data, potentially in… view at source ↗

**Figure 3.** Figure 3: Representative synthetic benchmark exemplars. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

**Figure 4.** Figure 4: PHATE visualizations of experts annotations. Some datasets can be classified as more than one category. Because expert annotation is performed visually, we computed 2-dimensional PHATE [16] and UMAP [17] embeddings of each preprocessed dataset. We chose this pair because PHATE is designed to preserve both local and global manifold structure and has been shown to retain geometry and preserve trajectorie… view at source ↗

**Figure 5.** Figure 5: Overview of the SCREEBTOWER algorithm. (1) Data colored by the discrete Morse function, with midpoint thresholds indicated. (2) Level sets defined by edges crossing each threshold. (3) The resulting Reeb graph. (4) The simplified Reeb graph after suppressing degree-2 subdivision vertices while preserving cycles. We provide a simple, but powerful method for identifying the data shape, from which the label c… view at source ↗

**Figure 6.** Figure 6: Web interface for data annotation and visualization. Key components include: (1) Dataset Name identifying the active dataset; (2) Algorithm Selection for switching between five PHATE plots (varying k) and a UMAP embedding; (3) Visualization Parameters displaying settings for the chosen algorithm; (4) Visualization Plot showing the point cloud embedding; and (5) Labeling Tools for non-exclusive labeling, de… view at source ↗

**Figure 7.** Figure 7: Representative synthetic examples comparing recovered graph structures across methods. [PITH_FULL_IMAGE:figures/full_fig_p019_7.png] view at source ↗

**Figure 8.** Figure 8: Representative real-world scRNA-seq datasets visualized using PHATE embeddings [PITH_FULL_IMAGE:figures/full_fig_p020_8.png] view at source ↗

read the original abstract

High-dimensional point cloud data arise across many scientific domains, especially single-cell biology. The shapes or topologies of these datasets determine the types of information that can be extracted. For example, clustered data supports cell-type identification, trajectory structures support transition analysis, and archetypal structures capture continua of cellular behaviors. Existing analysis pipelines often assume a specific shape. The standard Seurat pipeline combines UMAP visualization with Louvain clustering and therefore assumes clustered data, while tools such as Monocle and SPADE assume tree-like structures, and flow-based models such as MIOFlow and Conditional Flow Matching target trajectories. Choosing which pipeline to apply is therefore often left to bioinformaticians who visually inspect datasets before selecting an analysis strategy. With the rise of agentic AI scientists, automating shape detection is increasingly important for selecting downstream analysis pipelines. To address this problem, we introduce scShapeBench, a benchmark dataset for shape detection containing both synthetic and expert-annotated single-cell datasets. Synthetic datasets are sampled from ground-truth skeleton graphs with controlled variance. Real single-cell datasets are curated from diverse sources and annotated by experts into four categories: clusters, single trajectory, multi-branching, and archetypal. We additionally introduce scReebTower, a baseline method that uses diffusion geometry to extract Reeb graphs and connect visualization with pipeline selection. We provide topology-aware evaluation metrics and compare scReebTower against PAGA and Mapper on synthetic and real data. Our results indicate that scReebTower outperforms existing baselines. Overall, our contributions span benchmarks, evaluation metrics, and a baseline for automated shape detection in single-cell data.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

scShapeBench supplies a needed dedicated benchmark for shape detection in scRNA-seq, but the real-data outperformance rests on unvalidated expert labels.

read the letter

The core contribution here is scShapeBench itself: a new collection of synthetic point clouds drawn from known skeleton graphs plus real single-cell datasets labeled by experts into four discrete shape classes (clusters, single trajectory, multi-branching, archetypal). scReebTower adds a Reeb-graph baseline built on diffusion geometry that links visualization to downstream pipeline choice. That setup is genuinely new relative to the assumed-shape tools the abstract cites (Seurat, Monocle, PAGA, Mapper). The synthetic side is cleanly grounded because performance can be checked directly against the generating graphs, and the topology-aware metrics are a sensible way to score recovery of structure rather than just visual appeal. The motivation for automating shape detection also tracks with the rise of larger studies and agentic workflows. The soft spot is the real-data evaluation. Claims that scReebTower outperforms the baselines on expert-annotated datasets depend on those labels being accurate, consistent, and representative of the geometries that matter. The abstract supplies no inter-annotator agreement numbers, no sensitivity analysis to label noise, and no description of how ambiguous or mixed-topology cases were handled. If the annotations carry systematic bias or disagreement, the reported real-data advantage is not securely established even if the method itself is sound. Synthetic results can stand on their own, but the headline comparison mixes the two. This work is aimed at single-cell analysts who currently pick pipelines by eye and at groups building automated or large-scale analysis systems. A reader focused on manifold learning or topological data analysis in biology would get concrete value from the benchmark construction and metrics. The paper shows clear thinking on the problem setup and cites the relevant prior tools without obvious internal contradictions. It deserves peer review so referees can examine the full implementation details, code, and any additional validation on the annotations that the abstract omits.

Referee Report

3 major / 3 minor

Summary. The paper introduces scShapeBench, a benchmark dataset for shape detection in high-dimensional scRNA-seq data comprising synthetic point clouds sampled from ground-truth skeleton graphs with controlled variance and real datasets curated from diverse sources and expert-annotated into four discrete categories (clusters, single trajectory, multi-branching, archetypal). It proposes scReebTower, a baseline method that extracts Reeb graphs via diffusion geometry to automate shape detection and link visualization to downstream pipeline selection, and evaluates it against PAGA and Mapper using topology-aware metrics, claiming outperformance on both synthetic and real data.

Significance. If the results hold, the benchmark, topology-aware metrics, and scReebTower baseline would provide a useful standardized framework for automated geometry discovery in single-cell biology, helping select appropriate analysis tools (e.g., clustering vs. trajectory inference) amid diverse data structures. The combination of synthetic ground-truth evaluation and real-data testing is a strength, as is the explicit connection between Reeb graphs and pipeline choice.

major comments (3)

[Real data curation and annotation] Real-data evaluation section: the outperformance claim on expert-annotated datasets rests on the assumption that the four-category labels are accurate, consistent, and representative, yet no inter-annotator agreement statistics, sensitivity analysis to label flips, or protocol for handling ambiguous/mixed-topology datasets are reported. This is load-bearing for the headline result, since the topology-aware metrics are defined relative to these labels.
[Results and evaluation] Results section: the abstract asserts that scReebTower outperforms PAGA and Mapper on topology-aware metrics, but supplies no numerical values, confidence intervals, statistical tests, or details on how synthetic variance or annotation disagreements were quantified. Without these, the strength of the superiority claim cannot be assessed.
[scReebTower method] Method section: while scReebTower builds on standard diffusion geometry and Reeb graphs, the specific choices for diffusion operator construction, Reeb graph extraction parameters, and the mapping from graph to pipeline recommendation are not fully specified, limiting reproducibility and making it hard to isolate what drives any performance gain.

minor comments (3)

[Abstract] Abstract: the phrase 'controlled variance' for synthetic data is used without specifying the variance schedule or sampling procedure.
[References] Ensure all cited baselines (PAGA, Mapper, Seurat, Monocle, MIOFlow) have complete references.
[Figures] Figure captions for Reeb graph visualizations should include explicit legends explaining node/edge coloring and any scale information.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their constructive comments, which highlight important aspects for strengthening the manuscript. We address each major comment below and will revise the paper accordingly to improve clarity, reproducibility, and the robustness of our claims.

read point-by-point responses

Referee: [Real data curation and annotation] Real-data evaluation section: the outperformance claim on expert-annotated datasets rests on the assumption that the four-category labels are accurate, consistent, and representative, yet no inter-annotator agreement statistics, sensitivity analysis to label flips, or protocol for handling ambiguous/mixed-topology datasets are reported. This is load-bearing for the headline result, since the topology-aware metrics are defined relative to these labels.

Authors: We agree that validating the annotation process is essential for the real-data results. In the revised manuscript, we will report inter-annotator agreement statistics (Cohen's kappa) from the expert annotations, include a sensitivity analysis demonstrating metric stability under label perturbations, and detail our protocol for ambiguous cases (majority vote with exclusion of unresolved mixed-topology samples). These additions will directly address the load-bearing nature of the labels. revision: yes
Referee: [Results and evaluation] Results section: the abstract asserts that scReebTower outperforms PAGA and Mapper on topology-aware metrics, but supplies no numerical values, confidence intervals, statistical tests, or details on how synthetic variance or annotation disagreements were quantified. Without these, the strength of the superiority claim cannot be assessed.

Authors: We acknowledge that the abstract and results lack sufficient quantitative support. We will revise the abstract to include specific performance values with confidence intervals and add statistical tests (e.g., paired Wilcoxon tests) in the results section. We will also expand the evaluation protocol to explicitly describe how synthetic variance was controlled and how annotation disagreements were quantified and mitigated. revision: yes
Referee: [scReebTower method] Method section: while scReebTower builds on standard diffusion geometry and Reeb graphs, the specific choices for diffusion operator construction, Reeb graph extraction parameters, and the mapping from graph to pipeline recommendation are not fully specified, limiting reproducibility and making it hard to isolate what drives any performance gain.

Authors: We thank the referee for this observation on reproducibility. In the revision, we will fully specify the diffusion operator construction (kernel type, bandwidth selection via cross-validation), Reeb graph extraction parameters (level-set discretization and merging thresholds), and the exact rule-based mapping from extracted graph features to pipeline recommendations. We will also add pseudocode for the full pipeline. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper introduces scShapeBench (synthetic data from explicit ground-truth skeleton graphs plus expert-annotated real scRNA-seq data) and scReebTower (diffusion geometry plus Reeb graphs, standard techniques). It evaluates against external baselines (PAGA, Mapper) using topology-aware metrics on both synthetic and real data. No equation or step reduces by construction to a fitted parameter, self-definition, or load-bearing self-citation chain; the outperformance claim is an empirical comparison on independent inputs rather than a tautology.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 1 invented entities

The central claim rests on the assumption that the four shape categories capture biologically relevant geometries and that expert labels are reliable ground truth for real data.

axioms (2)

domain assumption Expert annotations of real single-cell datasets into clusters, single trajectory, multi-branching, and archetypal categories are accurate and consistent.
Used to create ground-truth labels for real-data evaluation.
domain assumption Synthetic data sampled from ground-truth skeleton graphs with controlled variance adequately models real single-cell variability.
Basis for controlled synthetic test cases.

invented entities (1)

scReebTower no independent evidence
purpose: Baseline method that extracts Reeb graphs via diffusion geometry to detect shape and select analysis pipelines.
Newly introduced component whose performance is the main empirical result.

pith-pipeline@v0.9.0 · 5668 in / 1374 out tokens · 40296 ms · 2026-05-14T21:17:41.406383+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Foundation/AlexanderDuality.lean alexander_duality_circle_linking unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

scReebTower relies on the diffusion geometry to extract Reeb graphs... Use the Fiedler vector ϕ1 as a Morse function for Reeb graph construction
IndisputableMonolith/Foundation/AbsoluteFloorClosure.lean absolute_floor_iff_bare_distinguishability unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

We provide topology-aware evaluation metrics... Wasserstein persistence similarity and graph edit distance

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

31 extracted references · 31 canonical work pages · 2 internal anchors

[1]

Spatial reconstruction of single-cell gene expression data.Nat Biotechnol, 33(5):495–502, April 2015

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

work page 2015
[2]

The single-cell transcriptional landscape of mammalian organogenesis.Nature, 566(7745):496–502, February 2019

Junyue Cao, Malte Spielmann, Xiaojie Qiu, Xingfan Huang, Daniel M Ibrahim, Andrew J Hill, Fan Zhang, Stefan Mundlos, Lena Christiansen, Frank J Steemers, Cole Trapnell, and Jay Shendure. The single-cell transcriptional landscape of mammalian organogenesis.Nature, 566(7745):496–502, February 2019

work page 2019
[3]

Visualization and cellular hierarchy inference of single-cell data using spade.Nature protocols, 11(7):1264–1279, 2016

Benedict Anchang, Tom DP Hart, Sean C Bendall, Peng Qiu, Zach Bjornson, Michael Linder- man, Garry P Nolan, and Sylvia K Plevritis. Visualization and cellular hierarchy inference of single-cell data using spade.Nature protocols, 11(7):1264–1279, 2016

work page 2016
[4]

Guillaume Huguet, D. S. Magruder, Alexander Tong, Oluwadamilola Fasina, Manik Kuchroo, Guy Wolf, and Smita Krishnaswamy. Manifold interpolating optimal-transport flows for trajectory inference, 2022

work page 2022
[5]

Trajecto- rynet: A dynamic optimal transport network for modeling cellular dynamics

Alexander Tong, Jessie Huang, Guy Wolf, David Van Dijk, and Smita Krishnaswamy. Trajecto- rynet: A dynamic optimal transport network for modeling cellular dynamics. InInternational conference on machine learning, pages 9526–9536. PMLR, 2020

work page 2020
[6]

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 Göttgens, Nikolaus Rajewsky, Lukas Simon, and Fabian J Theis. PAGA: graph abstraction reconciles clustering with trajectory inference through a topology preserving map of single cells. Genome Biol, 20(1):59, March 2019

work page 2019
[7]

Topological Methods for the Analysis of High Dimensional Data Sets and 3D Object Recognition

Gurjeet Singh, Facundo Memoli, and Gunnar Carlsson. Topological Methods for the Analysis of High Dimensional Data Sets and 3D Object Recognition. In M. Botsch, R. Pajarola, B. Chen, and M. Zwicker, editors,Eurographics Symposium on Point-Based Graphics. The Eurographics Association, 2007

work page 2007
[8]

Sur les points singuliers d’une forme de pfaff complètement intégrable ou d’une fonction numérique.Comptes Rendus de l’Académie des Sciences, 222:847–849, 1946

Georges Reeb. Sur les points singuliers d’une forme de pfaff complètement intégrable ou d’une fonction numérique.Comptes Rendus de l’Académie des Sciences, 222:847–849, 1946

work page 1946
[9]

Diffusion maps.Applied and computational harmonic analysis, 21(1):5–30, 2006

Ronald R Coifman and Stéphane Lafon. Diffusion maps.Applied and computational harmonic analysis, 21(1):5–30, 2006

work page 2006
[10]

Time-inhomogeneous diffusion geometry and topology, 2023

Guillaume Huguet, Alexander Tong, Bastian Rieck, Jessie Huang, Manik Kuchroo, Matthew Hirn, Guy Wolf, and Smita Krishnaswamy. Time-inhomogeneous diffusion geometry and topology, 2023

work page 2023
[11]

Moyle, Manik Kuchroo, David van Dijk, Kevin R

Nathan Brugnone, Alex Gonopolskiy, Mark W. Moyle, Manik Kuchroo, David van Dijk, Kevin R. Moon, Daniel Colon-Ramos, Guy Wolf, Matthew J. Hirn, and Smita Krishnaswamy. Coarse graining of data via inhomogeneous diffusion condensation. In2019 IEEE International Conference on Big Data (Big Data), page 2624–2633. IEEE, December 2019. 10

work page 2019
[12]

Cz cellxgene discover: a single-cell data platform for scalable exploration, analysis and modeling of aggregated data.Nucleic acids research, 53(D1):D886–D900, 2025

CZI Cell Science Program, Shibla Abdulla, Brian Aevermann, Pedro Assis, Seve Badajoz, Sidney M Bell, Emanuele Bezzi, Batuhan Cakir, Jim Chaffer, Signe Chambers, et al. Cz cellxgene discover: a single-cell data platform for scalable exploration, analysis and modeling of aggregated data.Nucleic acids research, 53(D1):D886–D900, 2025

work page 2025
[13]

Single cell portal: an interactive home for single-cell genomics data.BioRxiv, 2023

Leyla Tarhan, Jon Bistline, Jean Chang, Bryan Galloway, Emily Hanna, and Eric Weitz. Single cell portal: an interactive home for single-cell genomics data.BioRxiv, 2023

work page 2023
[14]

Expression atlas update: insights from sequencing data at both bulk and single cell level.Nucleic Acids Research, 52(D1):D107–D114, 2024

Nancy George, Silvie Fexova, Alfonso Munoz Fuentes, Pedro Madrigal, Yalan Bi, Haider Iqbal, Upendra Kumbham, Nadja Francesca Nolte, Lingyun Zhao, Anil S Thanki, et al. Expression atlas update: insights from sequencing data at both bulk and single cell level.Nucleic Acids Research, 52(D1):D107–D114, 2024

work page 2024
[15]

Scanpy: large-scale single-cell gene expression data analysis.Genome biology, 19(1):15, 2018

F Alexander Wolf, Philipp Angerer, and Fabian J Theis. Scanpy: large-scale single-cell gene expression data analysis.Genome biology, 19(1):15, 2018

work page 2018
[16]

Moon, David van Dijk, Zheng Wang, Scott Gigante, Daniel B

Kevin R. Moon, David van Dijk, Zheng Wang, Scott Gigante, Daniel B. Burkhardt, William S. Chen, Kristina Yim, Antonia van den Elzen, Matthew J. Hirn, Ronald R. Coifman, Natalia B. Ivanova, Guy Wolf, and Smita Krishnaswamy. Visualizing structure and transitions in high- dimensional biological data.Nature Biotechnology, 37(12):1482–1492, 2019

work page 2019
[17]

UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction

Leland McInnes, John Healy, and James Melville. Umap: Uniform manifold approximation and projection for dimension reduction.arXiv preprint arXiv:1802.03426, 2018

work page internal anchor Pith review Pith/arXiv arXiv 2018
[18]

Multiscale phate identifies multimodal signatures of covid-19.Nature biotechnology, 40(5):681–691, 2022

Manik Kuchroo, Jessie Huang, Patrick Wong, Jean-Christophe Grenier, Dennis Shung, Alexan- der Tong, Carolina Lucas, Jon Klein, Daniel B Burkhardt, Scott Gigante, et al. Multiscale phate identifies multimodal signatures of covid-19.Nature biotechnology, 40(5):681–691, 2022

work page 2022
[19]

Heterogenous populations of tissue-resident cd8+ t cells are generated in response to infection and malignancy

J Justin Milner, Clara Toma, Zhaoren He, Nadia S Kurd, Quynh P Nguyen, Bryan McDonald, Lauren Quezada, Christella E Widjaja, Deborah A Witherden, John T Crowl, et al. Heterogenous populations of tissue-resident cd8+ t cells are generated in response to infection and malignancy. Immunity, 52(5):808–824, 2020

work page 2020
[20]

Revealing dynamic temporal regulatory networks driving cancer cell state plasticity with neural ode-based optimal transport

Smita Krishnaswamy, Alex Tong, Manik Kuchroo, Shabarni Gupta, Aarthi Venkat, Beatriz San Juan, Laura Rangel, Brandon Zhu, John Lock, and Christine Chaffer. Revealing dynamic temporal regulatory networks driving cancer cell state plasticity with neural ode-based optimal transport. 2023

work page 2023
[21]

Aanet resolves a continuum of spatially localized cell states to unveil intratumoral heterogeneity

Aarthi Venkat, Scott E Youlten, Beatriz P San Juan, Carley A Purcell, Shabarni Gupta, Matthew Amodio, Daniel P Neumann, John G Lock, Anton E Westacott, Cerys S McCool, et al. Aanet resolves a continuum of spatially localized cell states to unveil intratumoral heterogeneity. Cancer Discovery, 15(10):2139–2165, 2025

work page 2025
[22]

Probability measures on the space of persistence diagrams.Inverse Problems, 27(12):124007, nov 2011

Yuriy Mileyko, Sayan Mukherjee, and John Harer. Probability measures on the space of persistence diagrams.Inverse Problems, 27(12):124007, nov 2011

work page 2011
[23]

Exploring network structure, dynamics, and function using networkx

Aric Hagberg, Pieter Swart, and Daniel Chult. Exploring network structure, dynamics, and function using networkx. 06 2008

work page 2008
[24]

An Exact Graph Edit Distance Algorithm for Solving Pattern Recognition Problems

Zeina Abu-Aisheh, Romain Raveaux, Jean-Yves Ramel, and Patrick Martineau. An Exact Graph Edit Distance Algorithm for Solving Pattern Recognition Problems. In4th International Conference on Pattern Recognition Applications and Methods 2015, Lisbon, Portugal, January 2015

work page 2015
[25]

Persistence images: A stable vector representation of persistent homology.Journal of Machine Learning Research, 18(8):1–35, 2017

Henry Adams, Tegan Emerson, Michael Kirby, Rachel Neville, Chris Peterson, Patrick Shipman, Sofya Chepushtanova, Eric Hanson, Francis Motta, and Lori Ziegelmeier. Persistence images: A stable vector representation of persistent homology.Journal of Machine Learning Research, 18(8):1–35, 2017

work page 2017
[26]

Comparing partitions.Journal of Classification, 2(1):193– 218, December 1985

Lawrence Hubert and Phipps Arabie. Comparing partitions.Journal of Classification, 2(1):193– 218, December 1985

work page 1985
[27]

A comparison of single-cell trajectory inference methods.Nature Biotechnology, 37(5):547–554, April 2019

Wouter Saelens, Robrecht Cannoodt, Helena Todorov, and Yvan Saeys. A comparison of single-cell trajectory inference methods.Nature Biotechnology, 37(5):547–554, April 2019. 11

work page 2019
[28]

Stability of persistence diagrams

David Cohen-Steiner, Herbert Edelsbrunner, and John Harer. Stability of persistence diagrams. volume 37, pages 263–271, 06 2005

work page 2005
[29]

Alexander Wolf, Florian Buettner, and Fabian J

Laleh Haghverdi, Maren Büttner, F. Alexander Wolf, Florian Buettner, and Fabian J. Theis. Diffusion pseudotime robustly reconstructs lineage branching.Nature Methods, 13(10):845–848, Oct 2016

work page 2016
[30]

How Powerful are Graph Neural Networks?

Keyulu Xu, Weihua Hu, Jure Leskovec, and Stefanie Jegelka. How powerful are graph neural networks?arXiv preprint arXiv:1810.00826, 2018

work page internal anchor Pith review Pith/arXiv arXiv 2018
[31]

Strategies for pre-training graph neural networks.arXiv preprint arXiv:1905.12265, 2019

Weihua Hu, Bowen Liu, Joseph Gomes, Marinka Zitnik, Percy Liang, Vijay Pande, and Jure Leskovec. Strategies for pre-training graph neural networks.arXiv preprint arXiv:1905.12265, 2019. 12 A Related Work Graph-based and topological structure recovery.Many methods for recovering structure from high-dimensional data begin by constructing a neighborhood grap...

work page arXiv 1905