SEAHORSE: A Serendipity Engine Assaying Heterogeneous Omics-Related Sampling Experiments

Adam Quackenbush, Jaya Kolluri, Rohan Varghese Biju, Saron Nhong, Derrick K. DeConti, John Quackenbush, Enakshi SahabioRxiv 2025 · DOI

Abstract

Large-scale, open-access data sets such as the Genotype Tissue Expression Project (GTEx) and The Cancer Genome Atlas (TCGA) include multi-omic data on large numbers of samples along with extensive clinical and phenotypic information. These datasets provide a unique opportunity to discover correlations among clinical and genomic data features that can lead to testable hypotheses and new discoveries. SEAHORSE (http://seahorse.networkmedicine.org/) is a web-based database and search tool for exploratory data analysis in which we have pre-computed statistical associations between available data elements. An easy-to-use user interface allows users to explore significant associations using tabulated summary statistics, data visualizations, and functional enrichment analyses (using RNA-seq data) for identified sets of genes. We describe the motivation and construction of SEAHORSE and demonstrate its utility by documenting several surprising association patterns observed across multiple tissues in GTEx and multiple different cancer types in TCGA.

Introduction

Scientific progress is often constrained by the "known unknowns"—incremental hypotheses based on existing literature. While this approach is foundational, it often fails to uncover "unknown unknowns," or unexpected relationships hidden within massive datasets that no researcher thought to investigate. Large-scale projects like the Genotype-Tissue Expression (GTEx) project and The Cancer Genome Atlas (TCGA) have generated vast amounts of multi-omic and phenotypic data. However, the computational burden of cross-referencing hundreds of clinical variables against tens of thousands of genes across dozens of tissues remains a significant barrier. SEAHORSE (Serendipity Engine Assaying Heterogeneous Omics-Related Sampling Experiments) addresses this by providing a pre-computed, searchable database designed to facilitate serendipitous discovery.

Methods

The authors developed a robust pipeline to systematically assay these massive datasets through a pre-computation strategy. Instead of calculating correlations on-the-fly, the system pre-calculates associations between all pairwise sets of phenotypic variables and gene expression levels. To maintain statistical rigor, specific tests were applied based on the data types involved in each comparison.

Statistical tests utilized in SEAHORSE based on data type pairings.

As illustrated in the figure above, categorical data comparisons utilize Chi-squared or Cramer’s V tests, while comparisons between categorical and continuous data (including gene expression) employ t-tests or ANOVA. Continuous-vs-continuous relationships, such as gene-gene or gene-phenotype correlations, are assessed via Pearson correlation. To ensure biological interpretability, the pipeline performs Gene Set Enrichment Analysis (GSEA) using the fgsea R package, linking gene expression patterns to KEGG, GO, and Reactome pathways.

Datasets

The tool currently leverages two primary pillars of modern genomics:

  1. GTEx v8: Data from 948 donors across 43 unique tissues, including 189 health-related variables (e.g., height, weight, race, type of death).
  2. TCGA: Over 20,000 primary tumor and normal samples across 33 cancer types, integrated via the Recount3 resource for uniformly processed RNA-seq data.

Key Insights and Results

SEAHORSE operates at a massive scale, having calculated and ranked 247,544,281 associations between phenotypes and gene expression levels, alongside over 415 million significant gene-gene correlations (using a cutoff of $r^2 > 0.7$).

The "Serendipity Factor"

The tool’s primary value lies in expanding the "Johari Window" of research by uncovering unexpected molecular links. For example, SEAHORSE revealed a striking correlation between human height and "Pathways in Cancer" across 22 different tissues. While epidemiological studies have long noted that taller individuals have a higher cancer risk, SEAHORSE provides a molecular map of the underlying gene expression changes that might explain this link. Additionally, height was significantly correlated with "cardiac muscle contraction" pathways in 19 tissues, offering a potential shared mechanism for height-associated heart conditions like atrial fibrillation.

Universal Aging Patterns

In the TCGA data, the tool identified that pathways involving cell adhesion and WNT signaling were significantly associated with age across more than 15 different cancer types. This suggests common biological mechanisms for why cancer prognosis often worsens with age, regardless of the specific tumor type.

Discussion and Future Work

The authors acknowledge the "multiple testing problem" inherent in running millions of statistical tests, which can lead to false positives. Consequently, SEAHORSE is positioned as a hypothesis-generation tool; findings should be validated in independent datasets.

Future developments for the platform include expanding to other massive databases like TOPMed, AACR GENIE, and the UK Biobank. The team also plans to incorporate other "omics" layers beyond gene expression, such as DNA methylation and somatic mutations, and integrate advanced network analysis tools to move from simple correlations to complex regulatory models.

Summary updated

Abstract

Large-scale, open-access data sets such as the Genotype Tissue Expression Project (GTEx) and The Cancer Genome Atlas (TCGA) include multi-omic data on large numbers of samples along with extensive clinical and phenotypic information. These datasets provide a unique opportunity to discover correlations among clinical and genomic data features that can lead to testable hypotheses and new discoveries. SEAHORSE (http://seahorse.networkmedicine.org/) is a web-based database and search tool for exploratory data analysis in which we have pre-computed statistical associations between available data elements. An easy-to-use user interface allows users to explore significant associations using tabulated summary statistics, data visualizations, and functional enrichment analyses (using RNA-seq data) for identified sets of genes. We describe the motivation and construction of SEAHORSE and demonstrate its utility by documenting several surprising association patterns observed across multiple tissues in GTEx and multiple different cancer types in TCGA.
BibTeX
@misc{quackenbush2025seahorse,
  author = {Adam Quackenbush and Jaya Kolluri and Rohan Varghese Biju and Saron Nhong and Derrick K. DeConti and John Quackenbush and Enakshi Saha},
  title = {{SEAHORSE: A Serendipity Engine Assaying Heterogeneous Omics-Related Sampling Experiments}},
  year = {2025},
  doi = {10.1101/2025.08.15.670514},
  howpublished = {bioRxiv},
}
Figure