arXiv Machine Learning

Correlation-Guided Flow Matching with Annealed Masking for Spatial Transcriptomics Generation

CorrFlow is a new generative framework for predicting spatial transcriptomics from histology images. It explicitly models gene-gene interactions using an annealed masked flow matching strategy and a gene graph‑regularized optimization that incorporates prior knowledge from STRING and data‑driven co‑expression from WGCNA. Across 12 datasets, CorrFlow outperforms existing methods in average PCC and HPCC, producing more biologically coherent ST predictions.

arXiv AI
1d ago

More Genes, Faster Inference: Efficient Large-Panel Spatial Transcriptomics Prediction from Histology

GeneFlow is a new framework that predicts large-panel spatial gene expression directly from histology images. It combines flow‑matching generative modeling with structured multimodal fusion, encoding both tissue morphology and cross‑gene interactions via histology and gene tokenizers. The method achieves state‑of‑the‑art performance on multiple spatial transcriptomics benchmarks, delivering up to 56.8% improvement in 1,000‑gene prediction while drastically reducing computational cost.

By Songming Zhang, Yuxiao Luo, Shifu Luo, Siran Chen, Shuquan Su, Jinyan Li
arXiv Machine Learning
Oct 28

Biology-Guided Context-Aware Residual Propagation for Spatial Gene Expression Prediction from Histology

The paper introduces BCARP, a Biology‑Guided Context‑Aware Residual Propagation framework that predicts spatial gene expression from H&E‑stained histology images. Rather than directly decoding expression from visual features, BCARP treats histology embeddings as morphological priors and iteratively refines spot representations in a transcriptomics‑oriented latent space, integrating nuclei‑guided local context and prototype‑guided global propagation. Evaluations on three spatial transcriptomics datasets and ten HEST‑1k cohorts show that BCARP outperforms existing baselines in mean correlation‑based metrics, demonstrating the effectiveness of biology‑guided local‑global residual propagation for histology‑to‑transcriptomics prediction.

By Vu Van An, Long D. Nguyen, Ho Viet Duc Luong, Binh P. Nguyen
arXiv Machine Learning
Sep 25

SpaFactor: Lightweight Spatial Context-Aware Gene Program Modeling for Histology-to-Transcriptomics Inference

SpaFactor is a lightweight framework that predicts spatial gene expression from hematoxylin and eosin images by fusing central spot visuals with multiscale neighborhood context. It uses a residual MLP to map tissue microenvironment to low‑dimensional latent gene programs, which are decoded into coordinated multi‑gene predictions. Across five public cohorts, SpaFactor outperforms existing methods, especially for spatially variable genes, and better recovers biologically organized spatial patterns.

By Shiting Ruan, Xitong Ling, Qiming He, Ziyou Yan, Huaitian Yuan, Tian Guan, Ying Xiao, Xu Guan, Yonghong He
arXiv Computer Vision
Nov 6

Preserving DEG Rankings for Gene Discovery in Histology-Based Spatial Gene Expression Prediction

The paper introduces image-based differential expression ranking (IDER), a new objective for predicting spatial gene expression from histology images that focuses on preserving the ranking of differentially expressed genes (DEGs) rather than merely reconstructing per-gene spatial profiles. IDER compares gene rankings derived from differential‑expression statistics and offers a differentiable loss that can be trained using morphology‑derived proxy contrasts without predefined biological labels. Experiments on public spatial transcriptomics datasets demonstrate that IDER improves agreement with measured DEG rankings and pathway‑enrichment overlap compared to conventional reconstruction objectives, including evaluations based on morphology and pathologist‑annotated tissue regions.

By Kaito Shiku, Kazuya Nishimura, Yasuhiro Kojima, Ryoma Bise
arXiv Computer Vision
6d ago

3D-Guided Scalable Flow Matching for Generating Volumetric Tissue Spatial Transcriptomics from Serial Histology

The paper introduces HoloTea, a 3D‑aware flow‑matching framework that imputes spot‑level gene expression on unprofiled histology sections using H&E images and neighboring profiled sections. By incorporating cross‑section context via a lightweight ControlNet and a zero‑inflated negative binomial prior, HoloTea preserves anatomical continuity along the z‑axis and scales efficiently with a global attention block. Experiments on three spatial transcriptomics datasets show that HoloTea outperforms both 2D and existing 3D baselines in expression accuracy and generalization.

By Mohammad Vali Sanian, Arshia Hemmat, Amirhossein Vahidi, Jonas Maaskola, Jimmy Tsz Hang Lee, Martin Prete, Stanislaw Makarchuk, Yeliz Demirci, Nana-Jane Chipampe, Muzlifah Haniffa, Omer Bayraktar, Lassi Paavolainen, Mohammad Lotfollahi
arXiv Machine Learning
Jun 3

Spatial Transcriptomics-Guided Alignment Enhances Molecular Profiling in Pathology Foundation Model

arXiv:2606. 03644v1 Announce Type: new Abstract: Comprehensive molecular profiling is essential for modern precision oncology but remains hindered by prohibitive costs, specimen exhaustion, and protracted turnaround times.

By Fengtao Zhou, Yingxue Xu, Zhengyu Zhang, Yihui Wang, Zhengrui Guo, Ling Liang, Jiabo Ma, Cheng Jin, Ziyi Liu, Huajun Zhou, Hongyi Wang, Du Cai, Chenglong Zhao, Xi Wang, Can Yang, Yu Wang, Wenbin Li, Feng Gao, Zhe Wang, Zhenhui Li, Xiuming Zhang, Li Liang, Hao Chen
arXiv Computer Vision
6d ago

Predicting Single-Cell Spatial Gene Expression from H&E using Pathology Foundation Models

The paper presents a method for predicting single‑cell spatial gene expression from hematoxylin and eosin (H&E) images by adapting pathology foundation models with domain‑informed, task‑specific strategies. Using paired H&E and Xenium 5K data, the approach improves mean gene‑wise Pearson correlation by 2.67× over a zero‑shot baseline and achieves state‑of‑the‑art performance, with an AUROC of 0.800 across the full gene panel. The study shows that predictive performance varies across biological contexts, achieving the highest correlation for spatially variable genes (r = 0.643), and demonstrates that H&E‑derived representations can recover biologically meaningful spatial expression patterns despite sparse single‑cell measurements.

By Ashley P. Tsang, Evan Liu, Conrad Foo, Bob Chen, Hassaan Maan, Donny Chan, Elisa Penna, Veronica Ibarra Lopez, Hector Corrada Bravo, Aicha BenTaieb, Lisa M. McGinnis, Runmin Wei, Kenneth T. Gao