{
  "id": 14984291,
  "title": "GEMS (GWAS Effect Mechanisms via Sequence-to-Function): Unveiling cell type-specific mechanisms for GWAS variants",
  "url": "https://urgent.news/2026/10/16/gems-gwas-effect-mechanisms-via-sequence-to-function-unveiling-cell",
  "topic": "science",
  "section": "Science",
  "published": "2026-10-16T00:00:00.000Z",
  "source": {
    "name": "bioRxiv",
    "slug": "biorxiv",
    "url": "https://www.biorxiv.org/content/10.64898/2026.10.14.759213v1?rss=1"
  },
  "original_language": "en",
  "account": "Genome-wide association studies (GWAS) have identified thousands of genetic loci associated with common diseases, however, the specific cell types and conditions through which these non-coding variants influence disease risk remain largely elusive. To address this knowledge gap, researchers have introduced an innovative framework called GEMS (GWAS Effect Mechanisms via Sequence-to-Function). This framework employs supervised classifiers trained on multimodal variant effect predictions derived from ATAC-seq and RNA-seq data. By leveraging the Borzoi model, which translates genomic sequences into functional outcomes, GEMS is then able to associate individual GWAS variants with the most relevant cell types via SHAP-based attribution.\n\nThe GEMS framework was applied to four distinct diseases: asthma, LDL cholesterol, multiple sclerosis, and type 2 diabetes. The results of this study underscore the framework's ability to prioritize cell types that are known to be relevant to the disease in question, and these findings align with the results obtained through stratified LD-score regression. Importantly, GEMS was able to associate a median of 58.7% of fine-mapped variants with at least one cell type through ATAC-seq analysis, and a higher 65.9% association via RNA-seq analysis. This suggests that GEMS extends cellular annotation beyond the chromatin accessibility peaks present in the single-cell atlas used for training the model.\n\nThe associations between ATAC-seq variants and specific cell types were found to be significantly enriched in cis-regulatory elements identified in independent experimental studies that were not incorporated into the model training process. Furthermore, this enrichment was observed to persist even after excluding variants that overlapped the training atlas peaks. This discovery indicates that the GEMS framework can uncover novel, cell-type-specific mechanisms underlying the association between GWAS variants and disease risk.\n\nOne particularly compelling outcome of the GEMS framework is its ability to categorize risk variants into distinct immune and CNS (central nervous system) modules for the case of multiple sclerosis. This hierarchical biclustering approach revealed that these risk variants are regulated by different transcription factor families, and these findings successfully recapitulate the dual nature of multiple sclerosis as both an immune-mediated and neurodegenerative disease. The integration of sequence-to-function predictions using GEMS thus provides a means to stratify phenotype-associated variants into mechanistically distinct, cell-type-specific regulatory programs.",
  "summary": "Genome-wide association studies (GWAS) have linked thousands of loci to common diseases, yet for non-coding variants, which constitute the vast majority of the signal, the cell types and states through which individual risk variants act remain largely unknown. Here, we develop GEMS (GWAS Effect Mechanisms via Sequence-to-Function), a framework that trains supervised classifiers on multimodal…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}