Pretrained gene representations transfer mean expression more broadly than spatial patterns in virtual spatial transcriptomics
- 1Johns Hopkins Bloomberg School of Public Health
Abstract
Virtual spatial transcriptomics models combine tissue images with pretrained gene representations to predict expression for unseen genes, but what these representations contribute is unclear. We separated their effects into overall gene-mean and spatial variation using held-out genes and individuals across four cohorts spanning three human brain regions and HER2-positive breast cancer. With fixed representations derived from regulatory DNA sequences and representations pretrained on single-cell transcriptomes, reductions in gene-mean error accounted for 91-96% of the total mean squared error reductions over matched random vectors. Mean-only models using gene representations without images retained 82-99% of the corresponding full-matrix correlation gain. Spatial gains were smaller, increased with expression variation in training tissue, and varied across cohorts and representations. Thus, pretrained gene representations broadly transfer mean expression but only selectively support spatial recovery, showing that gene-panel extension alone does not establish spatial recovery for new targets.
Related
- SpaCoEx: Sparse Gene Selection for Spatially Varying Co-expression in Spatial Transcriptomics
- NicheCorr: Identifying Niche-Specific Differential Gene Co-Expression in Spatial Transcriptomics
- Full-length transcriptomics and proteomics reveal how genome minimization reshapes gene expression in synthetic bacteria
- Comparative single-cell transcriptomics links gene expression divergence to Drosophila genital morphogenesis
- MorphoANOVA: Dissecting Histology-Gene Expression Relationships in Spatial Transcriptomics
BibTeX
@misc{chen2026pretrained,
title = {{Pretrained gene representations transfer mean expression more broadly than spatial patterns in virtual spatial transcriptomics}},
author = {T. Chen and S. C. Hicks},
howpublished = {bioRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.15.751768},
url = {https://www.biorxiv.org/content/10.64898/2026.09.15.751768v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Pretrained gene representations transfer mean expression more broadly than spatial patterns in virtual spatial transcriptomics
AU - T. Chen
AU - S. C. Hicks
PY - 2026
DA - 2026/09/15
PB - bioRxiv
DO - 10.64898/2026.09.15.751768
UR - https://www.biorxiv.org/content/10.64898/2026.09.15.751768v1
AB - Virtual spatial transcriptomics models combine tissue images with pretrained gene representations to predict expression for unseen genes, but what these representations contribute is unclear. We separated their effects into overall gene-mean and spatial variation using held-out genes and individuals across four cohorts spanning three human brain regions and HER2-positive breast cancer. With fixed representations derived from regulatory DNA sequences and representations pretrained on single-cell transcriptomes, reductions in gene-mean error accounted for 91-96% of the total mean squared error reductions over matched random vectors. Mean-only models using gene representations without images retained 82-99% of the corresponding full-matrix correlation gain. Spatial gains were smaller, increased with expression variation in training tissue, and varied across cohorts and representations. Thus, pretrained gene representations broadly transfer mean expression but only selectively support spatial recovery, showing that gene-panel extension alone does not establish spatial recovery for new targets.
ER -