Author Correction: Regulatory genomic circuitry of human disease loci by integrative epigenomics

Carles A. Boix, Benjamin T. James, Yongjin Park, Wouter Meuleman, Manolis KellisNature 2025 · DOI · OpenAlex

Introduction

In 2021, a significant study was published in Nature detailing the regulatory genomic circuitry of human disease loci through integrative epigenomics. This work aimed to bridge the gap between genetic variants identified in genome-wide association studies (GWAS) and their functional roles in tissue-specific regulatory elements, such as enhancers. Recently, the authors issued a formal correction to address two specific implementation errors in their bioinformatics pipeline that affected the statistical significance of their reported findings.

Identifying Statistical Implementation Errors

The correction stems from the discovery of two primary technical issues in the original analysis:

  1. Directionality of Enrichment Calculation: The original pipeline calculated the enrichment of enhancers within GWAS lead SNPs. Because enhancers often cluster spatially (within 2.5 kb), multiple enhancers were frequently counted as independent hits for a single SNP. This "upside-down" approach led to double-counting, which artificially inflated the statistical significance of the overlaps.
  2. Incomplete Multiple Hypothesis Testing Correction: The initial False Discovery Rate (FDR) procedure was applied only to local enhancer clusters. It failed to account for the global scale of testing across hundreds of distinct tissue-trait combinations, resulting in values that were closer to nominal P-values than true FDR-corrected significance.

Corrective Methodology

To rectify these errors, the authors re-analyzed the data using a more robust statistical framework. The primary change involved reversing the direction of the enrichment query: assessing the enrichment of GWAS SNPs within enhancers. In this configuration, a single SNP is counted only once if it overlaps with any enhancer in a given biosample, effectively neutralizing the spatial clustering bias. Additionally, the authors implemented a global FDR correction across all tissue-trait pairs to ensure rigorous multiple testing control.

Impact on Findings

The re-analysis led to a refinement of the study's quantitative results. Under a strict 1% global FDR threshold, the number of GWAS showing significant epigenomic enrichment was adjusted from 245 to 226. This correction necessitated updates to several key figures (Figures 3 and 4) and numerous extended data and supplementary figures.

Despite the reduction in the number of significant associations, the core biological conclusion remains intact: disease-associated SNPs are highly enriched in tissue-specific regulatory elements. The stability of 226 significant studies confirms the robustness of the primary "circuitry" model proposed in the original work.

Conclusion and Scientific Transparency

This correction highlights the critical importance of implementation details in large-scale genomic data integration. By providing updated figures and maintaining access to the original datasets for comparison, the authors demonstrate a commitment to scientific transparency and reproducible research. For the broader genomics community, this serves as a cautionary example of how spatial feature distribution and the scope of FDR correction can significantly influence statistical outcomes.

Summary updated

BibTeX
@inproceedings{boix2025author,
  author = {Carles A. Boix and Benjamin T. James and Yongjin Park and Wouter Meuleman and Manolis Kellis},
  title = {{Author Correction: Regulatory genomic circuitry of human disease loci by integrative epigenomics}},
  booktitle = {Nature},
  year = {2025},
  doi = {10.1038/s41586-025-09134-4},
}
Figure