Spatial Gene Expression Prediction

Spatial gene expression prediction aims to computationally infer gene expression levels within tissue samples using readily available data like histology images, thereby reducing reliance on expensive and complex spatial transcriptomics experiments. Current research heavily utilizes deep learning, employing architectures such as transformers, diffusion models, and graph neural networks to integrate information from multiple image resolutions and modalities (e.g., combining histology with existing spatial transcriptomic data). This field is significant because it promises to make spatial transcriptomics data more accessible, facilitating large-scale studies and potentially improving disease diagnosis and treatment by providing spatially resolved insights into gene expression patterns.