Efficient prediction of a spatial transcriptomics profile better characterizes breast cancer tissue sections without costly experimentation

TLDR

Spatial transcriptomics is an emerging technology that requires costly reagents and significant expertise, limiting the identification of transcriptional markers linked to histology. The study aims to show that predicting spatial gene expression in unmeasured regions can enhance biologists’ histological interpretations. The authors developed DeepSpaCE, a deep‑learning model for spatial gene clusters and expression, validated it on consecutive breast cancer tissue sections, and extended it to a semi‑supervised version that improves imputation accuracy using unlabeled histology images. Predicted SPARC expression uncovered a small tumor‑invasion region that raw data missed, and the method enables users to derive hidden histological characters from spatial transcriptomes and gene annotations, accelerating discoveries without additional experiments.

Abstract

Abstract Spatial transcriptomics is an emerging technology requiring costly reagents and considerable skills, limiting the identification of transcriptional markers related to histology. Here, we show that predicted spatial gene-expression in unmeasured regions and tissues can enhance biologists’ histological interpretations. We developed the Deep learning model for Spatial gene Clusters and Expression, DeepSpaCE, and confirmed its performance using the spatial-transcriptome profiles and immunohistochemistry images of consecutive human breast cancer tissue sections. For example, the predicted expression patterns of SPARC , an invasion marker, highlighted a small tumor-invasion region difficult to identify using raw spatial transcriptome data alone because of a lack of measurements. We further developed semi-supervised DeepSpaCE using unlabeled histology images and increased the imputation accuracy of consecutive sections, enhancing applicability for a small sample size. Our method enables users to derive hidden histological characters via spatial transcriptome and gene annotations, leading to accelerated biological discoveries without additional experiments.