Sang Lee

TACO detects presymptomatic Long COVID by combining causal inference with a foundation model, achieving higher precision with fewer features and no hyperparameter tuning.

This study identifies genetic factors underlying personality meta-traits stability and plasticity and their links to psychopathology using GWAS and structural equation modeling.

This paper reviews computational methods and datasets for understanding Long COVID's mechanisms and developing diagnostic and therapeutic strategies using multi-omics data.

This paper identifies 32 causal genes for Long COVID using a multi-omics framework combining TWMR and CT, revealing novel genetic factors and symptom subtypes.

The paper introduces R2ROC, a method to compare AUC values from polygenic scores efficiently, offering faster computation and direct mapping to R2 for better interpretation.

deepKin estimates relatedness with statistical inference and applies it to UK Biobank data, identifying significant relatives and geographic patterns.

This study examines cross-ancestry genetic architecture of cholesterol traits, finding significant genetic heterogeneity and identifying concordant SNPs with higher predictive ability.

Cross-trait Elastic Net mixing of PGS Catalog scores (PRSmix+) lifted mean PRS accuracy 1.72-fold in Europeans and 1.42-fold in South Asians over the best single score.

This study introduces a novel GxE PRS model that improves type 1 error control and statistical power for genetic risk prediction, with applications to quantitative and binary traits.

The paper examines how genetic and brain similarities predict childhood anthropometric and socioeconomic traits using a mixed model framework in the ABCD cohort.

This paper introduces a method to estimate cross-ancestry genetic correlation using individual-level data, finding significant genetic heterogeneity in obesity between African and European ancestry groups.

The paper develops statistical methods to estimate variance and covariance of R2 values for polygenic scores, enabling significance testing of predictive ability differences across samples and SNP types.

This paper introduces a new hyper-parameter, the scale factor, to improve the prediction accuracy of the HBLUP method in genetic evaluation using simulated and real cattle data.

The paper investigates gene-environment interactions for depressive symptoms using UK Biobank data, finding limited polygenic-environment interactions but significant residual-environment interactions.

The paper identifies genetic-by-environment interactions affecting diabetes-related metabolic traits using UK Biobank data, showing lifestyle impacts vary by genetic background.

The paper introduces a linear mixed model integrating genomic and exposomic data to improve phenotypic prediction for complex traits, showing increased variance explanation and prediction accuracy.

The paper introduces CORE GREML, a method to estimate covariance between random effects in LMMs for genomic analyses, showing improved performance in variance partitioning.

GxEsum estimates phenotypic variance from genome-wide GxE interactions using GWAS summary statistics, showing efficiency and accuracy on biobank-scale data.

The paper introduces a method to detect genotype-population interaction effects using ancestry principal components to analyze genetic heterogeneity across populations.

This study performs a cross-disorder GWAS meta-analysis of ADHD, ASD, OCD, and TS, identifying 297 genome-wide significant variants and 21 genes associated with shared genetic risk.

This paper investigates the genetic links between female reproductive traits and six psychiatric disorders using polygenetic risk scores from UK Biobank data.

RICOPILI is a Perl-based pipeline for efficient GWAS analysis, including quality control, imputation, and meta-analysis, with automated parallelization for rapid data processing.

This study uses a whole-genome approach to identify genetic and non-genetic variance components in cardiovascular risk influenced by lifestyle factors, showing their modulating effects and implications for CVD prevention.

This paper introduces a multivariate reaction norm model to estimate genotype-covariate correlation and interaction in complex trait analyses, showing improved performance over existing methods.

The paper finds a genetic association between schizophrenia risk and age at first birth in women, with different genetic correlations in younger and older age groups.

This study evaluates bias and accuracy in genomic breeding value prediction under selection scenarios using different statistical models and finds that SSGBLUP with proper scaling reduces bias and improves accuracy.

The paper develops a theoretical framework to estimate genomic prediction accuracy using reference populations with varying relationships to the target, showing that close relatives significantly improve accuracy.

This study investigates genetic mechanisms underlying sex bias in ADHD prevalence using genome-wide data and finds no mean polygenic burden difference across sexes but increased risk in female siblings and comorbid conditions.