Beyond benchmarking: an expert-guided consensus approach to spatially aware clustering
Journal article, 2026
Spatial omics technologies have revolutionized the study of tissue architecture and cellular heterogeneity by integrating molecular profiles with spatial localization. In spatially resolved transcriptomics, delineating higher-order anatomical structures is critical for understanding how cellular organization affects function. However, the reliability of current benchmarks of spatially aware clustering (SAC) methods is undermined by their narrow focus on Visium and brain tissue datasets and the incorrect interpretation of manual annotation as ground truth. Here we present SACCELERATOR, a community-driven, extensible framework that standardizes data formatting, method integration and metric evaluation, enabling rapid inclusion of new methods and datasets. Our analysis revealed substantial limitations in the generalizability and reproducibility of SAC methods and shows that anatomical labels commonly used as ground truths are often biased, error prone and unsuitable for benchmarking. Rather than ranking methods, we propose a consensus-guided workflow where descriptive spatial metrics highlight high-entropy regions of method disagreement, enabling targeted feedback for tissue experts. Applied to brain and cancer datasets, this approach uncovered biologically meaningful patterns overlooked by individual SAC methods and manual annotations, highlighting the need for iterative, expert-in-the-loop evaluation.
Author
Jieran Sun
Centre Hospitalier Universitaire Vaudois
Kirti Biharie
Leiden University
Delft University of Technology
Peiying Cai
University of Zürich
Niklas Müller-Bötticher
Berliner Institut für Gesundheitsforschung
Paul Kiessling
Psychosomatics
Meghan A. Turner
Allen Institute for Brain Science
Søren Helweg Dam
University of Copenhagen
Technical University of Denmark (DTU)
Florian Heyl
German Cancer Research Center (DKFZ)
The German Human Genome-Phenome Archive (GHGA)
Sarusan Kathirchelvan
University of Zürich
Martin Emons
University of Zürich
Samuel Gunz
University of Zürich
Sven Twardziok
Berliner Institut für Gesundheitsforschung
Amin El-Heliebi
Medical University of Graz
Martin Zacharias
Medical University of Graz
Roland Eils
Berliner Institut für Gesundheitsforschung
M. J. T. Reinders
Delft University of Technology
Raphael Gottardo
Centre Hospitalier Universitaire Vaudois
Christoph Kuppe
Psychosomatics
Brian Long
Allen Institute for Brain Science
Ahmed Mahfouz
Leiden University
Delft University of Technology
Mark D. Robinson
University of Zürich
Naveed Ishaque
Berliner Institut für Gesundheitsforschung
Nature Methods
1548-7091 (ISSN) 1548-7105 (eISSN)
Vol. In PressSubject Categories (SSIF 2025)
Computer Sciences
DOI
10.1038/s41592-026-03194-8
PubMed
42637977