Image based gene ranking improves gene discovery in tissue analysis
Preserving DEG Rankings for Gene Discovery in Histology-Based Spatial Gene Expression Prediction
Computer Vision and Pattern Recognition
Summary
Spatial transcriptomics lets scientists see where genes are active in tissue samples, but it can be expensive and slow. This paper shows how predicting gene activity from tissue images can better preserve which genes stand out between different tissue regions. The authors develop a new way to train models that focuses on keeping the right gene ranking for group differences, rather than just matching gene activity levels. This method helps discover important genes linked to biological features or diseases more accurately from images alone.
What this means in practice
- •For clinical pathology teams: Use improved gene ranking from tissue images to identify disease relevant genes without needing full gene expression data.
- •For bioinformatics developers: Build better tools to predict biologically meaningful gene differences from histology images to assist gene discovery workflows.
Authors
Kaito Shiku, Kazuya Nishimura, Yasuhiro Kojima, Ryoma Bise
Abstract
Predicting spatial gene expression from histology images could scale spatial transcriptomics (ST) to image-only cohorts, but conventional histology-based ST prediction is trained and evaluated mainly by per-gene spatial-profile reconstruction. This objective is misaligned with a key downstream use of ST: differentially expressed gene (DEG) discovery, where genes are ranked for a biological or morphology-defined contrast by evidence of between-group expression differences. We formulate image-based differential expression ranking (IDER), which asks whether predicted expression profiles preserve the contrast-specific ranked gene list obtained from measured profiles. IDER compares gene rankings induced by differential-expression statistics, rather than raw expression magnitudes or per-gene spatial correlations. We further introduce a differentiable IDER objective that aligns these statistics across genes and can be trained with morphology-derived proxy contrasts without predefined biological group labels. Experiments on public ST datasets show improved DEG-ranking agreement and pathway-enrichment overlap over conventional reconstruction objectives, including morphology-derived and pathologist-annotated tissue-region evaluations.