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

TL;DR

MoLF is a novel generative model that predicts spatial gene expression across multiple cancer types from histology, leveraging a mixture-of-experts architecture for improved pan-cancer performance.

Contribution

Introduces MoLF, a mixture-of-latent-flow model with a conditional flow matching objective, enabling effective pan-cancer histogenomic prediction and zero-shot cross-species generalization.

Findings

MoLF outperforms existing models on pan-cancer benchmarks.

MoLF achieves zero-shot generalization to cross-species data.

MoLF establishes a new state-of-the-art in spatial gene expression prediction.

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.