Matthew Stephens

The paper introduces a new enrichment analysis method using GWAS summary statistics to identify trait-associated genes, pathways, and tissues across 31 human phenotypes, uncovering previously unreported associations.

This paper develops a method for genotyping polyploid organisms using NGS data, incorporating error modeling and Mendelian segregation, and applies it to hexaploid sweet potatoes.

This paper introduces a grade of membership model to infer and visualize DNA damage patterns in ancient DNA, distinguishing ancient from modern samples and contamination.

CorShrink is an Empirical Bayes method for estimating correlations with missing data, showing competitive performance on gene expression and word vector similarity tasks.

The paper examines H3K9me3 marking of transposable elements in human and chimpanzee induced pluripotent stem cells, finding limited species-specificity and no strong link to gene expression divergence.

This paper introduces a Bayesian framework for large-scale multiple regression using GWAS summary statistics, enabling heritability estimation and association detection without individual-level data.

The paper applies grade of membership models to RNA-seq data to identify and summarize biological structure and heterogeneity in both bulk and single-cell datasets.

The paper develops a method to estimate the time to the common ancestor of a beneficial allele using a hidden Markov model and applies it to human population data.

This study identifies 7,273 new human coding sequences using ribosome footprint profiling and validates their functionality through selective constraint and translation correlation analysis.

The paper introduces a method to visualize spatial population structure using effective migration surfaces, identifying gene flow barriers from genetic data.

The paper introduces msCentipede, a model that improves transcription factor binding site inference by better modeling DNase I cleavage pattern heterogeneity and spatial structure.

This study performs a genome-wide association analysis of LDL and IDL subfractions and their response to statins in 1868 individuals, identifying strong genetic associations and demonstrating increased power with multivariate analysis.

This study maps eQTL and analyzes gene expression variation in wild baboons, finding strong genetic contributions and similarities with human data.

The paper develops fast variational Bayesian algorithms for inferring population structure in large SNP datasets, achieving speed improvements over STRUCTURE and comparable accuracy to ADMIXTURE.

The paper investigates how epigenetic marks contribute to gene expression differences between primate species using lymphoblastoid cell lines.