FLAG: Foundation model representation with Latent diffusion Alignment via Graph for spatial gene expression prediction

2026-05-18

AI Summary

Existing methods for spatial gene expression prediction often overlook the interplay between gene co-regulation and spatial topology, limiting their ability to model complex gene–space joint relationships. This work proposes FLAG, a novel framework that reformulates the task as structured distribution modeling. FLAG integrates spatial graph neural networks with a Gene Foundation Model (GFM) alignment mechanism, enabling simultaneous preservation of spatial topological consistency and fidelity of inter-gene regulatory relationships during latent diffusion-based generation. To address the “curse of gene dimensionality,” the study introduces new evaluation metrics—Gene Structural Correlation and Spatial Structural Correlation—that significantly enhance the quality of structural relationship reconstruction. FLAG maintains competitive performance on conventional accuracy measures such as PCC and MSE while achieving more biologically realistic modeling of joint gene–spatial expression patterns.

Abstract

Predicting spatial gene expression from routine H\&E enables large-scale molecular profiling, yet current models treat this as isolated pointwise tasks, thereby overlooking essential biological structures like gene coordination and spatial distribution. To preserve these relationships, we introduce \textbf{FLAG}, a diffusion-based framework that redefines this task as structured distribution modeling. At the same time, we identify the critical \textbf{Gene Dimension Curse}, where joint modeling gene expression and their spatial interactions fail in high-dimensional spaces, and FLAG solves this challenge by integrating a spatial graph encoder for topological consistency and utilizing Gene Foundation Model (GFM) alignment for gene-gene fidelity in the generation process. To rigorously assess model performance, we propose a set of novel structural evaluation metrics, including Gene Structural Correlation (\textbf{GSC}) and Spatial Structural Correlation (\textbf{SSC}). Our experiments demonstrate that FLAG is highly competitive in traditional accuracy (PCC/MSE) while achieving significantly enhanced structural fidelity in capturing both gene-gene and gene-spatial relationships. The code is available at https://github.com/darkflash03/FLAG.

Problem

• spatial gene expression
• gene coordination
• spatial distribution
• Gene Dimension Curse
• structured distribution modeling

Innovation

• Latent Diffusion
• Spatial Gene Expression
• Graph Encoder
• Foundation Model Alignment
• Structural Evaluation Metrics