Mengyu He

This study benchmarks 12 methods for differential composition analysis in scRNA-seq data, finding LOCOM performs best in controlling false discovery rate and maintaining power.

This paper introduces gemGAT, a graph attention network for gene expression imputation across tissues, showing improved performance over regression-based methods on GTEx and ADNI data.

TestNet infers microbial networks with FDR control, handling clustered and unclustered samples by generating null replicates that account for compositional effects and clustering.

This paper introduces SMRmix, a mixed effect model for integrating multiple microbiome datasets to identify consistent community-level associations with outcomes.

LOCOM2 improves differential abundance analysis in microbiome studies by controlling error rates and handling challenges like relative abundance data and unbalanced designs.

This study identifies causal relationships between gut microbiome, metabolic dysfunction-associated fatty liver disease (MAFLD), and blood metabolites using Mendelian randomization.

MIDASim simulates realistic microbiome data by capturing taxon correlations, sparsity, and compositionality, outperforming existing methods in accuracy and efficiency.