Dajiang J. Liu

This paper introduces a supervised contrastive learning framework for multi-omics prediction models, showing improved accuracy over single-omics approaches in type 2 diabetes and high-fat diet studies.

The paper integrates proteomics data from Olink and SomaScan to identify pQTLs and develop a resource for transcriptome and proteome-wide association studies, with applications to inflammatory bowel disease.

This study identifies B cell-intrinsic and -extrinsic mechanisms by which IRF7 promotes spontaneous autoimmune germinal center and plasma cell responses, autoantibody production, and SLE development.

This paper develops a method to improve Mendelian randomization analysis by integrating GWAS results across ancestries, enhancing accuracy and power in discovering causal genes for autoimmune diseases.

The paper develops polygenic risk scores to stratify lung cancer and smoking cessation risk, showing significant odds ratios in diverse populations and planning clinical trials to evaluate interventions.

The paper investigates the genetic impact of the X-chromosome and sex differences in the human brain, identifying 72 genome-wide significant trait-locus pairs linked to brain-related disorders.

The paper investigates genetic contributions to skin color variation in a Caribbean population with Native American, African, and European ancestry, identifying specific alleles and ancestry effects on pigmentation.

MB-SupCon integrates microbiome and metabolomics data to improve prediction accuracy in microbiome-only datasets, achieving high accuracy in predicting insulin resistance, sex, and race.

This study identifies 35 genes associated with circulating lipid levels using rare coding variants from 170,000 exomes across multiple ancestries.

The paper develops KIRCLE to genotype KIR genes from whole exome sequencing data and analyzes its associations with HLA alleles and clinical traits in TCGA and UK Biobank.

MetaPrism enables joint analysis of taxa and gene abundances in metagenomic data to identify potential biomarkers for therapeutic response prediction.

MB-GAN generates synthetic microbiome data that mimic real datasets, preserving key statistical properties for method development.

The paper uses multivariate genomic analysis of 1.5 million people to identify genetic loci related to addiction, antisocial behavior, and health, finding over 500 loci enriched for brain-expressed genes.

The paper analyzes rare coding variants in smoking and drinking behavior using exome chip data, finding that they contribute to phenotypic variation but require larger samples for individual variant associations.

This study identifies 16 causal coding variants in type 2 diabetes using fine-mapping and shows that one third of signals are false leads.

The paper develops a method to handle missing data in genetic meta-analysis, identifying novel low-frequency variants in nicotine dependence genes with improved accuracy.

The paper develops methods to analyze multi-allelic variants in large sequence datasets, showing improved power and unbiased estimates in association tests.