SPATNIC enables H&E-grounded cancer-cell identification across spatial and single-cell transcriptomic modalities in colorectal cancer
- 1Division of Translational Informatics, Exploratory Oncology Research & Clinical Trial Center, National Cancer Center, Kashiwa, Japan; Department of Computationa
Abstract
Tumors contain diverse non-cancer cells that constitute the tumor microenvironment; thus, identifying which cells are cancerous is a foundational and critical first step in cancer research. However, to our knowledge, no transcriptome-based classifier has been trained using pathologist-defined annotations from matched H&E images at single-cell resolution. Moreover, few tools can be shared across diverse platforms. Here, we present SPATNIC, a variational autoencoder that classifies individual cells as cancer or normal, trained on Xenium cells labeled by a pathologist review of matched H&E images. It matches or exceeds the performance of existing tools in primary colorectal tumors and liver and lung metastases. SPATNIC generalizes zero-shot across Visium, Visium HD, CosMx, Xenium, MERSCOPE, and single-cell RNA sequencing. In public single-cell atlases, it identified a rare POU2F3 tuft-like cancer stem cell population missed by existing annotations and associated with worse survival. In rectal cancer profiled before and after chemoradiotherapy, it separated cancer from normal epithelium in tissue, thus enabling per-patient screening of antibody-drug conjugate (ADC) targets.
Related
- Artificial Intelligence for Colorectal Cancer Diagnosis: A Comprehensive Survey of Modalities, Frameworks, and Clinical Tasks
- Integrated Transcriptomic Analysis Identifies ENO3 as a Candidate Lactate Metabolism‐Associated Biomarker in Colorectal Cancer Metastasis
- Stage-Stratified Explainable Machine Learning Identifies a Reproducible Transcriptomic Classifier with Stage-Dependent Feature Importance in Colorectal Cancer: A Multi-Cohort Study with Saudi Population Contextualisation
- An interpretable whole-blood transcriptomic model for Parkinson disease identification and molecular profiling
- Identification of candidate selenium-associated transcriptomic biomarkers in chronic obstructive pulmonary disease: an integrated bioinformatics and expression validation study
BibTeX
@misc{sakai2026spatnic,
title = {{SPATNIC enables H&E-grounded cancer-cell identification across spatial and single-cell transcriptomic modalities in colorectal cancer}},
author = {S. A. Sakai and R. Nomura and M. Ohira and K. Kurihara and S. Chi and Y. Matsubara and Y. Morinaga and T. Kato and N. Takeda-Miyata and M. Kojima and S. Sakashita and H. Ebi and R. Yamaguchi and E. Oki and T. Fujisawa and H. Bando and A. Suzuki and Y. Suzuki and K. Tsuchihara and S.-I. Kageyama and R. Yamashita},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.21.760612},
url = {https://www.biorxiv.org/content/10.64898/2026.10.21.760612v1},
note = {CC BY-NC},
}
RIS
TY - UNPB
TI - SPATNIC enables H&E-grounded cancer-cell identification across spatial and single-cell transcriptomic modalities in colorectal cancer
AU - S. A. Sakai
AU - R. Nomura
AU - M. Ohira
AU - K. Kurihara
AU - S. Chi
AU - Y. Matsubara
AU - Y. Morinaga
AU - T. Kato
AU - N. Takeda-Miyata
AU - M. Kojima
AU - S. Sakashita
AU - H. Ebi
AU - R. Yamaguchi
AU - E. Oki
AU - T. Fujisawa
AU - H. Bando
AU - A. Suzuki
AU - Y. Suzuki
AU - K. Tsuchihara
AU - S.-I. Kageyama
AU - R. Yamashita
PY - 2026
DA - 2026/10/21
PB - bioRxiv
DO - 10.64898/2026.10.21.760612
UR - https://www.biorxiv.org/content/10.64898/2026.10.21.760612v1
AB - Tumors contain diverse non-cancer cells that constitute the tumor microenvironment; thus, identifying which cells are cancerous is a foundational and critical first step in cancer research. However, to our knowledge, no transcriptome-based classifier has been trained using pathologist-defined annotations from matched H&E images at single-cell resolution. Moreover, few tools can be shared across diverse platforms. Here, we present SPATNIC, a variational autoencoder that classifies individual cells as cancer or normal, trained on Xenium cells labeled by a pathologist review of matched H&E images. It matches or exceeds the performance of existing tools in primary colorectal tumors and liver and lung metastases. SPATNIC generalizes zero-shot across Visium, Visium HD, CosMx, Xenium, MERSCOPE, and single-cell RNA sequencing. In public single-cell atlases, it identified a rare POU2F3 tuft-like cancer stem cell population missed by existing annotations and associated with worse survival. In rectal cancer profiled before and after chemoradiotherapy, it separated cancer from normal epithelium in tissue, thus enabling per-patient screening of antibody-drug conjugate (ADC) targets.
ER -