Research

Top

Research

Spatial transcriptomics

It models the gene expression matrix Y as a function of latent factors Z through a factor analysis model. Importantly, SpatialPCA builds a kernel matrix ? using the location information L to explicitly model the spatial correlation structure in the latent factors Z across tissue locations.

SpatialPCA is a spatially aware dimension reduction method that takes both gene expression Y and location information L as inputs.

Lulu Shang, Xiang Zhou
Nature Communications, 2022

Analyzing spatial transcriptomics data is computationally challenging, as the data are often noisy and display substantial spatial correlation across tissue locations. Here, we develop a spatially aware dimension reduction method, SpatialPCA, that can extract low dimensional representations of the spatial transcriptomics data with biological signal and preserved spatial correlation structure.

For a cell type of interest, Celina examines one gene at a time and takes as input the gene expression vector y, the cell type proportion (for spot-resolution data), or cell type indicator (for single-cell resolution data) vector x for the cell type of interest, any necessary covariates (?), and location information which is used for calculating the kernel matrix K. It relies on a spatially varying coefficient model to relate the gene��s spatial expression pattern to the cell type proportions across tissue locations and uses random effects to disentangle the cell type specific expression ?, which is further decomposed into two additional components: the spatial component ??, and the non-spatial component ??.

Celina is a method for identifying ct-SVGs for both single-cell resolution (top left) and spot-resolution (top right) spatial transcriptomics.

Lulu Shang*, Peijun Wu*, Xiang Zhou
Nature Communications, 2025

An essential task in spatial transcriptomics is identifying spatially variable genes (SVGs). Here, we present Celina, a statistical method for systematically detecting cell type-specific SVGs (ct-SVGs), a subset of SVGs exhibiting distinct spatial expression patterns within specific cell types.

Rank order the counts in reference data (1st, 5th, 10th, etc.) and use this to rank order the counts in synthetic data. SRTsim assigns the simulated counts to the locations in the synthetic data in a way that preserves the spatial expression pattern observed in the reference data.

Step 3 of the SRTsim schematic, assigning generated gene expression counts to the locations in the synthetic data.

Jiaqiang Zhu*, Lulu Shang*, Xiang Zhou
Genome Biology, 2023

Spatially resolved transcriptomics (SRT)-specific computational methods are often developed, tested, validated and evaluated in silico using simulated data. We present SRTsim, an SRT-specific simulator for scalable, reproducible and realistic SRT simulations. SRTsim not only maintains various expression characteristics of SRT data but also preserves spatial patterns.

Integrating genome-wide association studies (GWAS) with single cell

From left to middle: we calculate gene level effect measurements from GWAS summary statistics for the trait of interest and treat them as an m dimensional vector of outcomes. From right to middle: we infer the tissue specific gene co-expression network for each tissue/cell type using the PANDA software, which requires the tissue-specific gene expression matrix, existing protein-protein interaction network information, as well as existing transcription factor and gene binding information as input. For the trait of interest, we examine one tissue at a time and we model the gene-level effect measurements for the trait as a function of the gene co-expression matrix using a covariance regression network model. We infer parameters in the model through composite likelihood. We calculate the maximum composite likelihood for each tissue and eventually rank tissues by the corresponding log likelihoods.

Schematic of CoCoNet for identifying trait relevant tissue/cell type using information from gene co-expression networks

Lulu Shang, Jennifer A Smith, Xiang Zhou
PLOS Genetics, 2020

Understanding the biological functions of identified single-nucleotide polymorphism (SNP) associations requires identifying disease/trait relevant tissues or cell types. We developed a network method, CoCoNet, to incorporate tissue-specific gene co-expression networks constructed from single cell RNA sequencing studies with genome-wide association studies (GWAS) data for trait-tissue inference.

Quantitative trait locus (QTL) analysis

a�Cf P-values for the colocalizing eQTL (blue) and meQTL (red) are plotted against physical position. The genes in the tested eGene and meCpG pairs are plotted under the p-value figures. The first row is the test gene location, and the second row are other genes inside the given range. Associations were tested by two-sided Wald tests based on a sample size of n?=?961 for cis-meQTLs and n?=?1032 for cis-eQTLs.

Examples of top six colocalized eGene-meCpG pairs

Lulu Shang, Wei Zhao, Yi Zhe Wang, Zheng Li, Jerome J Choi, Minjung Kho, Thomas H Mosley, Sharon LR Kardia, Jennifer A Smith, Xiang Zhou
Nature Communications, 2023

Most existing meQTL mapping studies have focused on individuals of European ancestry and are underrepresented in other populations, with a particular absence of large studies in populations with African ancestry. We fill this critical knowledge gap by performing a large-scale cis-meQTL mapping study in 961 African Americans from the Genetic Epidemiology Network of Arteriopathy (GENOA) study.

(A and D) Histogram shows the number of eGenes (y axis) that harbor different numbers of independent eQTLs (x axis) in African Americans (A) or European Americans (D). We displayed eGenes that harbor up to nine independent eQTLs, with the detailed number of eQTLs listed above each bar. A large fraction of eGenes harbor a small number of independent eQTLs. (B and E) The proportion of variance (PVE) in gene expression levels explained by SNPs are higher for eGenes that harbor a larger number of independent eQTLs in African Americans (B) or European Americans (E). (C and F) Density plot shows the distance from eQTL to the transcription start site (TSS) of the corresponding eGene. The density plot is stratified by the number of eQTLs: eGenes with one independent eQTL is colored in red; eGenes with two independent eQTL is colored in green; eGenes with three independent eQTL is colored in blue; eGenes with four or more independent eQTL is colored in purple. Dashed lines represent the median distance between eQTL and TSS in the four stratified groups in African Americans (C) or European Americans (F).

Characterization of the Conditional eQTLs

Lulu Shang, Jennifer A Smith, Wei Zhao, Minjung Kho, Stephen T Turner, Thomas H Mosley, Sharon LR Kardia, Xiang Zhou
The American Journal of Human Genetics, 2020

Most existing expression quantitative trait locus (eQTL) mapping studies have been focused on individuals of European ancestry and are underrepresented in other populations including populations with African ancestry. We fill this critical knowledge gap by performing a large-scale in-depth eQTL mapping study on 1,032 African Americans (AA) and 801 European Americans (EA) in the GENOA cohort.

�鶹ӳ��

Research

Spatial transcriptomics

Integrating genome-wide association studies (GWAS) with single cell

Quantitative trait locus (QTL) analysis

Give Now

Research Areas

Subscribe