---
title: "Pretrained gene representations transfer mean expression more than spatial patterns in virtual spa"
id: "biorxiv-10-pretrained-gene-representations-transfer-mean-expression-more-broadly-than"
canonical_url: "https://medichelpline.com/clinical-feed/biorxiv-10-pretrained-gene-representations-transfer-mean-expression-more-broadly-than"
content_type: "clinical_feed_article"
specialty: "Oncology"
source_name: "bioRxiv (Biomedical Preprints)"
source_url: "https://www.biorxiv.org/content/10.64898/2026.09.15.751768v1?rss=1"
published_at: "2026-09-18T12:00:00.000Z"
evidence_level: "Verified Feed"
license: "CC-BY-NC-4.0 / Informational Use"
---
# Pretrained gene representations transfer mean expression more than spatial patterns in virtual spa
## Provenance & Clinical Metadata
- **Canonical URL:** https://medichelpline.com/clinical-feed/biorxiv-10-pretrained-gene-representations-transfer-mean-expression-more-broadly-than
- **Specialty:** [Oncology](https://medichelpline.com/clinical-feed/oncology.md)
- **Primary Source:** bioRxiv (Biomedical Preprints)
- **Source URL:** [Original Journal Publication](https://www.biorxiv.org/content/10.64898/2026.09.15.751768v1?rss=1)
- **Published At:** 2026-09-18T12:00:00.000Z
- **Evidence Rating:** Verified Feed
## Executive GIST (TL;DR)
- Models that combine tissue images with **pretrained gene representations** are evaluated for their ability to predict expression for held-out genes. - The authors decomposed held-out-gene prediction performance into two components: recovery of a gene's **mean expression** across tissue locations and recovery of its **spatial variation**. - Analyses covered four cohorts spanning three human brain regions and **HER2-positive breast cancer**, with held-out genes assessed in held-out individuals. - For spatial predictors using fixed representations from **Decima** or **scGPT**, reductions in gene-mean error explained more than 91% of the reduction in mean squared error versus matched random vectors. - Independently fitted mean-only models that used the same pretrained representations but no tissue images retained 90–99% of the corresponding gain in full-matrix correlation. - Improvements in recovering spatial patterns were smaller on average, varied by cohort and by the pretrained representation used, and tended to increase when expression variation was larger in the training tissue. - The findings indicate that pretrained gene representations broadly transfer information about **mean expression**, but only selectively improve recovery of **spatial expression patterns**. - The authors conclude that cross-gene generalization in virtual spatial transcriptomics comprises at least two distinct capabilities—mean-level transfer and spatial-pattern recovery—rather than a single unified skill.
## Clinical Analysis & Structured Key Points
Pretrained gene representations transfer mean expression more broadly than spatial patterns in virtual spatial transcriptomics | bioRxiv Skip to main content New Results Pretrained gene representations transfer mean expression more broadly than spatial patterns in virtual spatial transcriptomics View ORCID Profile Tingjun Chen , View ORCID Profile Stephanie C Hicks doi: https://doi.org/10.64898/2026.09.15.751768 Tingjun Chen Johns Hopkins Bloomberg School of Public Health Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Tingjun Chen Stephanie C Hicks Johns Hopkins Bloomberg School of Public Health Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Stephanie C Hicks For correspondence: shicks19{at}jhu.edu Abstract Info/History Metrics Supplementary material Data/Code Preview PDF Abstract Models that combine tissue images with pretrained gene representations aim to predict spatial expression for genes not used to fit the downstream predictor. Yet success on held-out genes can reflect two capabilities: estimating a gene's mean expression across tissue locations and recovering its spatial variation. Across four cohorts spanning three human brain regions and HER2-positive breast cancer, we evaluated held-out genes in held-out individuals and separated these components. For spatial predictors using fixed gene representations from Decima or scGPT, reductions in gene-mean error accounted for more than 91% of the reduction in mean squared error relative to matched random vectors. Independently fitted mean-only models using the same representations but no tissue images retained 90-99% of the corresponding gain in full-matrix correlation. Spatial gains were smaller on average, increased with expression variation in training tissue and differed across cohorts and representations. Across these settings, pretrained gene representations broadly transferred mean expression but selectively improved spatial recovery, showing that cross-gene generalization in virtual spatial transcriptomics is not a single capability. Competing Interest Statement The authors have declared no competing interest. Footnotes https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE264692 https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE307403 https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE307586 https://doi.org/10.5281/zenodo.17089020 https://doi.org/10.5281/zenodo.4751624 Funder Information Declared Johns Hopkins University, https://ror.org/00za53h95 Copyright The copyright holder for this preprint is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY 4.0 International license . Back to top Previous Next Posted September 18, 2026. Download PDF Supplementary Material Data/Code Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Pretrained gene representations transfer mean expression more broadly than spatial patterns in virtual spatial transcriptomics Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Pretrained gene representations transfer mean expression more broadly than spatial patterns in virtual spatial transcriptomics Tingjun Chen , Stephanie C Hicks bioRxiv 2026.09.15.751768; doi: https://doi.org/10.64898/2026.09.15.751768 Share This Article: Copy Citation Tools Pretrained gene representations transfer mean expression more broadly than spatial patterns in virtual spatial transcriptomics Tingjun Chen , Stephanie C Hicks bioRxiv 2026.09.15.751768; doi: https://doi.org/10.64898/2026.09.15.751768 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Areas All Articles Animal Behavior and Cognition (8010) Biochemistry (18733) Bioengineering (14876) Bioinformatics (44392) Biophysics (22588) Cancer Biology (19713) Cell Biology (26896) Clinical Trials (138) Developmental Biology (13961) Ecology (21001) Epidemiology (2067) Evolutionary Biology (25436) Genetics (16162) Genomics (23498) Immunology (18693) Microbiology (42451) Molecular Biology (18054) Neuroscience (93419) Paleontology (700) Pathology (2977) Pharmacology and Toxicology (5092) Physiology (8111) Plant Biology (15997) Scientific Communication and Education (2095) Synthetic Biology (4559) Systems Biology (10233) Zoology (2389)
## Related Clinical Research

- [Automated OCR extraction of genomic biomarkers (Oncotype DX) to reduce oncology data latency](https://medichelpline.com/clinical-feed/pubmed-42763830.md) (DOI: 10.1007/s10552-026-02252-y)
- [Inflammatory Biomarkers and Risk of Postmenopausal ER-Positive Breast Cancer](https://medichelpline.com/clinical-feed/british-journal-of-cancer-0-inflammatory-biomarkers-and-risk-of-postmenopausal-oestrogen-receptor-positive.md)
- [Pemetrexed Stabilizes BRCA1 and Enhances Radiosensitivity in Triple‑Negative Breast Cancer](https://medichelpline.com/clinical-feed/pubmed-42757486.md) (DOI: 10.3892/or.2026.9194)
- [Educating Patients on ‘Asian Glow’ and Alcohol Risks in East Asian Populations](https://medichelpline.com/clinical-feed/stat-news-0-opinion-doctors-should-educate-patients-about-asian-glow-and-drinking-risks.md)
- [Ultrasound Radiomics for Preoperative Prediction of Central Lymph Node Metastasis in Papillary Thy](https://medichelpline.com/clinical-feed/frontiers-in-immunology-15-noninvasive-immune-inflammatory-profiling-by-ultrasound-radiomics-for.md)

## Navigation
- [← Back to Oncology Feed](https://medichelpline.com/clinical-feed/oncology.md)
- [← All Clinical Specialties](https://medichelpline.com/clinical-feed.md)
## Medical & Regulatory Disclaimer

> [!CAUTION]
> MedicHelpline content is structured for research, educational, and professional discovery purposes. It does not constitute individual medical advice, clinical diagnosis, or treatment recommendations.
> Always verify dosing, contraindications, and regulatory alerts against official product labeling and primary regulatory sources before clinical decision-making.