Hao Cheng

This paper improves causal variant identification by genome-wide fine-mapping, showing it outperforms existing methods and identifies new causal variants for complex traits.

This paper introduces SBayesAPP, a Bayesian model that estimates annotation-stratified pleiotropy and co-polygenicity between complex traits using GWAS summary statistics, revealing cell-type-specific genetic correlations.

This study integrates longitudinal metabolite profiles into genomic prediction models for pig traits, showing improved prediction accuracy depending on the trait and timepoint of metabolite sampling.

This paper introduces end-to-end genomic prediction to directly predict images and text from genetic markers, showing comparable accuracy to traditional numeric phenotype-based methods.

This study develops a microbiome-enabled genomic selection method that improves prediction accuracy for nitrogen-related traits in maize by integrating host SNPs and rhizobiome ASVs.

This paper introduces a homomorphic encryption method for genotypes and phenotypes to enable secure collaborative genomic analyses while preserving data confidentiality.

This paper develops DeepGCF, a deep learning method to predict functional conservation between human and pig using epigenome and transcriptome data, showing improved performance and insights into regulatory and heritability patterns.

The paper introduces a neural network framework for genomic evaluations that integrates pedigree, genotypes, and phenotypes, showing improved prediction performance over traditional methods.

The paper introduces MegaBayesianAlphabet, a Bayesian regression framework for genome-wide prediction and association studies, showing improved accuracy in genetic value prediction and marker discovery using high-dimensional phenotypic data.

The paper introduces NN-LMM, a method that extends linear mixed models to multi-layer neural networks for genomic prediction using intermediate omics data, showing improved performance over existing approaches.

This study develops a low-cost greenhouse-based high-throughput phenotyping platform to assess maize growth under plant growth-promoting bacteria inoculation, showing moderate to high genomic heritability and correlation with manual measurements.

The paper develops LPChoose, a linear programming approach to optimize sequencing resource allocation in genotyped livestock populations for efficient haplotype coverage.

MegaLMM improves genetic value prediction accuracy by analyzing thousands of traits simultaneously in genomic studies.

This paper introduces NN-Bayes, a Bayesian neural network method for genome-wide prediction and association studies, showing improved performance over linear models in simulations and real data.

This paper introduces SEM-BayesC{Pi}, a multi-trait Bayesian regression method for GWAS that incorporates causal structures, showing improved genotype-phenotype mapping compared to multi-trait BayesC{Pi}.

The paper derives an exact distribution of linkage disequilibrium (LD) over generations in a finite population, accounting for mutation, selection, and minor allele frequency filtering.

The paper identifies issues with cross-validation in multi-trait genomic prediction and proposes three solutions to improve model selection.

This paper develops and evaluates broad-based multi-trait Bayesian regression methods for genomic prediction, showing improved accuracy over single-trait and restricted multi-trait approaches.