Causal inference for multiple risk factors and diseases from genomics data

Nature Communications, Published online: 25 August 2026; doi:10.1038/s41467-026-76877-7 Mendelian randomization aims to identify causal relationships from genetic data, but results can be biased by confounding. Here, the authors develop CI-GWAS, a statistical framework that improves causal inference and shows that many suspected risk factor-disease associations in UK Biobank may reflect unmeasured confounding.