MoLF: Mixture-of-Latent-Flow for Pan-Cancer Spatial Gene Expression Prediction from Histology

2026-02-02

AI Summary

Existing approaches are often confined to single-cancer modeling, limiting their ability to leverage shared biological mechanisms across cancer types and suffering from performance degradation under data scarcity. To address these challenges, this work proposes MoLF—a generative model that integrates Mixture-of-Experts (MoE) with Conditional Flow Matching (CFM) to map histopathological images into a latent space of gene expression, enabling pan-cancer spatial transcriptomics prediction. MoLF is the first to incorporate an MoE architecture within a flow matching framework, effectively disentangling cross-tissue expression patterns and demonstrating zero-shot generalization across species. Experimental results show that MoLF outperforms both specialized and foundation models on pan-cancer benchmarks, establishing a new state of the art.

Technology Category

• Machine Learning: Mixture of Experts (MoE)
• Computer Vision: Multi-modal Vision
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: ML for personalized search and recommendations
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Inferring spatial transcriptomics (ST) from histology enables scalable histogenomic profiling, yet current methods are largely restricted to single-tissue models. This fragmentation fails to leverage biological principles shared across cancer types and hinders application to data-scarce scenarios. While pan-cancer training offers a solution, the resulting heterogeneity challenges monolithic architectures. To bridge this gap, we introduce MoLF (Mixture-of-Latent-Flow), a generative model for pan-cancer histogenomic prediction. MoLF leverages a conditional Flow Matching objective to map noise to the gene latent manifold, parameterized by a Mixture-of-Experts (MoE) velocity field. By dynamically routing inputs to specialized sub-networks, this architecture effectively decouples the optimization of diverse tissue patterns. Our experiments demonstrate that MoLF establishes a new state-of-the-art, consistently outperforming both specialized and foundation model baselines on pan-cancer benchmarks. Furthermore, MoLF exhibits zero-shot generalization to cross-species data, suggesting it captures fundamental, conserved histo-molecular mechanisms.

Problem

• spatial transcriptomics
• pan-cancer
• histology
• data scarcity
• model heterogeneity

Innovation

• Mixture-of-Experts
• Flow Matching
• Pan-Cancer
• Spatial Transcriptomics
• Zero-shot Generalization