30-Second Takeaway
- Persistent EBV DNA burden is heritable, enriched in immune regulatory regions, and modulated by MHC class II variation.
- Gene–environment and polygenic context reshape effect sizes, selection, and the clinical performance of variant- and polygenic-based risk tools.
- Deep learning and multi-omics causal inference sharpen regulatory variant interpretation and causal gene/protein maps for T2D and cardiometabolic disease.
Week ending January 31, 2026
Genomic context, environment, and molecular intermediates refining risk and mechanism in complex disease
Host genetic variation shapes persistent EBV DNA and multisystem disease associations
Reanalysis of WGS from 490,560 UK Biobank and 245,394 All of Us participants quantified persistent EBV DNA in blood from off-target reads. Higher EBV DNA levels associated with respiratory, autoimmune, neurologic, and cardiovascular diseases, linking chronic viral burden to diverse clinical phenotypes. GWAS showed heritability enrichment in immune regulatory regions and protein-altering variants in 148 genes influencing EBV DNA persistence. Single-cell and pathway analyses implicated variable antigen processing in B cells and antigen-presenting cells as primary determinants of persistence. HLA class II variation and predicted epitope presentation emerged as key modulators, illustrating how host genetics structures long-term EBV control.
Cross-population atlas reveals pervasive gene–environment interactions shaping complex traits
This compendium analyzed 440,210 European and Japanese individuals with replication in 539,794 diverse participants to map gene–environment interactions. Decomposing age, sex, and lifestyle contributions revealed reverse causality from disease-driven dietary changes and clarified interaction etiologies. Genome-wide interaction analyses recovered missing heritability and altered trait–trait relationships, directly affecting polygenic prediction accuracy and cross-population portability. Single-cell and omics-level analyses uncovered aging-related shifts in regulatory pathways and multiple sex-discordant lipid metabolism effects, informing drug target evaluation.
AlphaGenome delivers megabase-scale regulatory predictions for variant interpretation
AlphaGenome is a unified deep learning model taking 1 Mb of DNA sequence to predict thousands of functional genomic tracks at single-base resolution. Predicted modalities include gene expression, transcription initiation, chromatin accessibility, histone marks, transcription factor binding, chromatin contacts, and splicing features. Trained on human and mouse genomes, AlphaGenome matched or exceeded leading external methods in 25 of 26 variant effect benchmarks. The model mechanistically recapitulated clinically relevant variants near TAL1 and is released with tools for generating regulatory tracks and variant scores from sequence.
Treatment-linked DNA methylation mediates cardiometabolic late effects after childhood cancer
In 2,938 childhood cancer survivors from the St. Jude Lifetime Cohort, an EWAS identified 1,893 PBMC CpG sites associated with cardiometabolic risk factors. Five CpGs near CPT1A and LMNA associated with all evaluated cardiometabolic traits, suggesting shared epigenetic pathways of risk. Mediation analyses identified 24 CpGs that partially mediated relationships between prior genotoxic treatments and cardiometabolic outcomes, implicating inflammatory and metabolic pathways. Cg20370568, a cis-eQTM for ANTXR2, mediated 20% of the effect of body-trunk radiotherapy on abnormal glucose, highlighting a specific mechanistic biomarker.
References
Numbered in order of appearance. Click any reference to view details.
Additional Reads
Optional additional studies from this edition.