Andy W. Dahl

This paper introduces REGENA, a method to disentangle gene-environment interaction mechanisms, finding that half of significant GxE signals are scale-dependent and effect heterogeneity drives most remaining GxE heritability.

The paper evaluates non-linearities in phenotypic similarity versus genetic relatedness by analyzing three processes: imperfect tagging, assortative mating, and shared environments, using UK Biobank and All of Us data.

The paper shows that phenotype scale choice affects GxE signals in biobank data, with log scaling revealing meaningful effects and explaining a significant portion of PGSxSex interactions.

The paper shows genotype- and phenotype-based genetic instruments explain largely independent components of liability to psychiatric disorders, supported by theoretical analysis and simulations.

This study identifies synergistic sets of genetic risk variants in psychiatric disorders, suggesting subtypes and reinterpreting shared genetic effects.

This paper identifies how outcome scaling and endogenous treatment effects create spurious gene-environment interactions and proposes a diagnostic to detect these artifacts.

The paper constructs a cell-type-resolved chromatin accessibility atlas in the human intestine and identifies regulatory programs and genetic variants associated with Crohn's disease.

This study quantifies the heritability of drug response to statins, metformin, warfarin, and methotrexate in UK Biobank participants and identifies genetic modifiers of drug effects and side effects.

The paper evaluates the impact of non-random mating and environmental transmission on genetic methods and introduces a simulation tool to improve robustness.

This paper introduces CTMM, a model to detect cell type-specific interindividual variation in scRNA-seq data, finding that most transcriptome-wide variability between donors is differentiation stage-specific.

The paper introduces a variance components method to estimate gene-environment interaction heritability in complex traits using UK Biobank data, finding significant GxE effects for several trait-environment pairs.

AutoComplete improves phenotype imputation in biobank data, enhancing GWAS power and genetic discovery.

This paper introduces EFA, a model that factorizes polygenic epistasis to improve prediction and uncover biological pathways in complex traits.

The paper introduces PA-FGRS, a method to estimate genetic liability from family records, and shows it improves MDD classification and GWAS power when combined with genotypes.

This paper shows that cross-trait assortative mating explains genetic correlations between traits, suggesting many reported genetic similarities may be overestimated.

The paper investigates how allele frequency differences of causal variants impact PRS portability between European and African populations, finding they significantly reduce accuracy.

This paper improves genetic studies of MDD by integrating shallow and deep phenotypes, enhancing GWAS power and PRS accuracy while preserving specificity.

GBAT improves detection of trans-gene regulation by addressing multiple testing, read-mapping biases, and confounders in eQTL analysis.

The paper tests gene expression prediction models' performance across populations and finds poor generalizability to African Americans compared to European Americans.

The paper discusses the interpretation of negative heritability estimates and suggests they may reflect real biological features rather than just statistical noise.

The paper introduces a model for structured polygenic epistasis called Coordinated Interaction and tests it using the UK Biobank, finding evidence in 14 of 26 traits.

This paper introduces CelFiE, a method to estimate cell type contributions to cfDNA, showing it can detect rare cell types and find increased skeletal muscle cfDNA in ALS patients.

This paper introduces GxEMM, a mixed model for gene-environment interactions, showing it improves heritability estimates and reveals stress reduces heritability of major depression.

This paper introduces Reverse GWAS (RGWAS) to identify and validate phenotypic subtypes using genetics, finding three novel metabolic subtypes and showing differential treatment effects.

The paper identifies a bias in using principal components (PCs) for confounder correction in genomic studies, showing that it can lead to false positives, especially for large genetic effects.

The paper introduces ADGLM to model population variance structure, showing it biases genetic association studies and identifies novel genotype-variance associations.

The paper derives statistical properties of random-effects models for genetic heritability, showing small bias and exploring model behavior under violated assumptions.