# scQTLtools: An R package for single-cell eQTL analysis.
## Introduction
Expression quantitative trait loci (eQTL) analysis links variations in gene
expression levels to genotypes. scQTLtools is designed to identify genetic
variants that influence gene expression at the single-cell level, and can
also visualize the results. Our package includes data preprocessing and
multiple visualization options, providing researchers with a powerful tool for
exploring sc-eQTLs within specific cellular contexts.
## Citation
If you find this tool useful, please cite:
------------------------------------------------------------------------
***[https://github.com/XFWuCN/scQTLtools](https://github.com/XFWuCN/scQTLtools)***
***[https://bioconductor.org/packages/3.21/bioc/html/scQTLtools.html](https://bioconductor.org/packages/3.21/bioc/html/scQTLtools.html)***
***[https://bioconductor.org/packages/devel/bioc/html/scQTLtools.html](https://bioconductor.org/packages/devel/bioc/html/scQTLtools.html)***
------------------------------------------------------------------------
## Installation
```{r, eval = FALSE}
if (!require("BiocManager"))
install.packages("BiocManager")
BiocManager::install("scQTLtools")
```
## Overview of the package
The functions in scQTLtools can be categorized into data input, data
pre-process, sc-eQTL calling and visualization modules. The functions and their
brief descriptions are summarized below.
### General Workflow
Each module is summarized as shown below.
***[Overview](vignettes/Overview.png)***
scQTLtools requires two key input data: a single-cell gene expression dataset
and a corresponding SNP genotype matrix. The single-cell gene expression
dataset can be either a gene expression matrix or an object such as a Seurat v4
object or a Bioconductor SingleCellExperiment object. The input genotype matrix
should follow a 0/1/2/3 encoding scheme: 1 for homozygous reference genotype, 2
for homozygous alternative genotype, 3 for heterozygous genotype, and 0 for
missing values. Moreover, scQTLtools can support a simplified 0/1/2
encoding scheme, where 2 denotes a non-reference genotype.
Additionally, the package includes functionality to normalize the raw
single-cell gene expression matrix, and filter SNP–gene pairs. After
pre-process, scQTLtools implements the `callQTL()` function to identify
sc-eQTLs.
Moreover, visualization at the single-cell level demonstrates the specificity
of eQTLs across distinct cell types or cellular states.
## Comparison and advantages compared to similar works
We compared scQTLtools to other packages with similar functionality, including
eQTLsingle, SCeQTL, MatrixEQTL, and iBMQ, as shown in the table below.
***[Comparison](vignettes/Comparison.png)***
Among these tools, scQTLtools stands out for its comprehensive features:
(1) scQTLtools accepts SingleCellExperiment objects and Seurat objects as
input data formats, which are particularly beneficial for users working with
single-cell RNA-seq data, and promote the interoperability with the current
Bioconductor ecosystem.
(2) scQTLtools supports both binary and three-class genotype encodings,
enhancing its applicability across different genetic studies.
(3) scQTLtools offers extensive data pre-processing capabilities, including
quality control filtering for SNPs and genes, and normalization of raw gene
expression matrix. Users can also customize the genomic window for defining
SNP–gene pairs, enabling flexible cis-eQTL discovery.
(4) scQTLtools provides three commonly used statistical models, which cater to
various data distributions and analysis needs. This diversity allows users to
select the most appropriate model for their specific dataset.
(5) scQTLtools includes a wide range of visualization tools, which facilitate
detailed exploration and interpretation of eQTL results at the single-cell
level.
(6) scQTLtools supports grouping by both cell type and cell state, which is
crucial for analyzing the nuanced effects of genetic variants on gene
expression within heterogeneous cell populations.
Overall, scQTLtools offers a comprehensive suite of features that enhance the
analysis and interpretation of eQTLs.
## Required input files
The input file requires a single-cell gene expression dataset and a
corresponding SNP genotype matrix. The single-cell gene expression dataset can
be either a gene expression matrix or an object such as a Seurat v4 object or a Bioconductor SingleCellExperiment object.
- gene expression matrix: describes gene expressions, each row represents a
gene and each column represents a cell.
- Seurat object/SingleCellExperiment object: a Seurat/SingleCellExperiment
object.
- SNP genotype matrix: A matrix where each row represents a variant and each
column a cell, using a 0/1/2/3 encoding scheme.
The columns (cells) of the genotype matrix should correspond to the columns
(cells) of the gene expression matrix.
**Example**
```{r input, message=FALSE}
library(scQTLtools)
# gene expression matrix
data(GeneData)
# SeuratObject
data(Seurat_obj)
# load the genotype data
data(SNPData)
data(SNPData2)
```
# Create eQTL object
The createQTLObject class is an R object designed to store data related to eQTL
analysis, encompassing data lists, result data frames, and slots for
biClassify, species, and group information.
**Example**
```{r createObject_matrix, message=FALSE}
eqtl_matrix <- createQTLObject(
snpMatrix = SNPData,
genedata = GeneData,
biClassify = FALSE,
species = 'human',
group = NULL)
```
Users can set `biClassify` to TRUE to change the genotype encoding mode.
**Example**
```{r createObject_matrix_bi, message=FALSE}
eqtl_matrix_bi <- createQTLObject(
snpMatrix = SNPData,
genedata = GeneData,
biClassify = TRUE,
species = 'human',
group = NULL)
```
Users can use Seurat object instead of gene expression matrix.
**Example**
```{r createObject_seuratobject, message=FALSE}
eqtl_seurat <- createQTLObject(
snpMatrix = SNPData2,
genedata = Seurat_obj,
biClassify = FALSE,
species = 'human',
group = "celltype")
```
Users can also take SingleCellExperiment object as input.
**Example**
```{r createObject_sceobject, message=FALSE}
# Create a SingleCellExperiment object
library(SingleCellExperiment)
sce <- SingleCellExperiment(assays = list(counts = GeneData))
eqtl_sce <- createQTLObject(
snpMatrix = SNPData,
genedata = sce,
biClassify = FALSE,
species = 'human',
group = NULL)
```
# Normalize the raw gene expression matrix
Use `normalizeGene()` to normalize the raw gene expression matrix.
**Example**
```{r Normalize_matrix, message=FALSE}
eqtl_matrix <- normalizeGene(
eQTLObject = eqtl_matrix,
method = "logNormalize")
```
```{r Normalize_sceobject, message=FALSE}
eqtl_sce <- normalizeGene(
eQTLObject = eqtl_sce,
method = "logNormalize")
```
# Identify the valid SNP–gene pairs
Here we use `filterGeneSNP()` to filter SNP–gene pairs.
**Example**
```{r filter_matrix, message=FALSE}
eqtl_matrix <- filterGeneSNP(
eQTLObject = eqtl_matrix,
snpNumOfCellsPercent = 2,
expressionMin = 0,
expressionNumOfCellsPercent = 2)
```
```{r filter_seuratobject, message=FALSE}
eqtl_seurat <- filterGeneSNP(
eQTLObject = eqtl_seurat,
snpNumOfCellsPercent = 2,
expressionMin = 0,
expressionNumOfCellsPercent = 2)
```
```{r filter_sceobject, message=FALSE}
eqtl_sce <- filterGeneSNP(
eQTLObject = eqtl_sce,
snpNumOfCellsPercent = 2,
expressionMin = 0,
expressionNumOfCellsPercent = 2)
```
# Identify single-cell eQTLs
Here we use `callQTL()` to perform single cell eQTL analysis.
**Example**
```{r callQTL1_matrix, message=FALSE}
eqtl1_matrix <- callQTL(
eQTLObject = eqtl_matrix,
gene_ids = NULL,
downstream = NULL,
upstream = NULL,
gene_mart = NULL,
snp_mart = NULL,
pAdjustMethod = "bonferroni",
useModel = "linear",
pAdjustThreshold = 0.05,
logfcThreshold = 0.1)
```
```{r callQTL1_seuratobject, message=FALSE}
eqtl1_seurat <- callQTL(
eQTLObject = eqtl_seurat,
gene_ids = NULL,
downstream = NULL,
upstream = NULL,
gene_mart = NULL,
snp_mart = NULL,
pAdjustMethod = "bonferroni",
useModel = "linear",
pAdjustThreshold = 0.05,
logfcThreshold = 0.025)
```
```{r callQTL1_sceobject, message=FALSE}
eqtl1_sce <- callQTL(
eQTLObject = eqtl_sce,
gene_ids = NULL,
downstream = NULL,
upstream = NULL,
gene_mart = NULL,
snp_mart = NULL,
pAdjustMethod = "bonferroni",
useModel = "linear",
pAdjustThreshold = 0.05,
logfcThreshold = 0.025)
```
Users can use the parameter `gene_ids` to select one or several genes of
interest for identifying sc-eQTLs.
**Example**
```{r callQTL2_matrix, message=FALSE}
eqtl2_matrix <- callQTL(
eQTLObject = eqtl_matrix,
gene_ids = c("CNN2",
"RNF113A",
"SH3GL1",
"INTS13",
"PLAU"),
downstream = NULL,
upstream = NULL,
gene_mart = NULL,
snp_mart = NULL,
pAdjustMethod = "bonferroni",
useModel = "poisson",
pAdjustThreshold = 0.05,
logfcThreshold = 0.1)
```
Users can also set `upstream` and `downstream` to specify SNPs proximal to the
gene in the genome.
**Example**
```{r callQTL3_matrix, message=FALSE}
eqtl3_matrix <- callQTL(
eQTLObject = eqtl_matrix,
gene_ids = NULL,
downstream = -9e7,
upstream = 2e8,
gene_mart = NULL,
snp_mart = NULL,
pAdjustMethod = "bonferroni",
useModel = "poisson",
pAdjustThreshold = 0.05,
logfcThreshold = 0.05)
```
# Visualize the results.
Use `visualizeQTL()` to visualize sc-eQTL results. Four plot types are
supported: "histplot", "violin", "boxplot", or "QTLplot".
**Example**
violin plot:
```{r visualizeQTL_matrix, message=FALSE}
visualizeQTL(
eQTLObject = eqtl1_matrix,
SNPid = "1:632647",
Geneid = "RPS27",
groupName = NULL,
plottype = "QTLplot",
removeoutlier = TRUE)
```
QTLplot:
By incorporating cell annotation from Seurat or SingleCellExperiment objects,
scQTLtools can reveal cell-type-specific patterns of eQTL effects.
```{r visualizeQTL_seuratobject, message=FALSE}
visualizeQTL(
eQTLObject = eqtl1_seurat,
SNPid = "1:632647",
Geneid = "RPS27",
groupName = NULL,
plottype = "QTLplot",
removeoutlier = TRUE)
```
In addition, the parameter `groupName` is used to specify a particular
single-cell group of interest.
```{r visualizeQTL_seuratobject_groupName, message=FALSE}
visualizeQTL(
eQTLObject = eqtl1_seurat,
SNPid = "1:632647",
Geneid = "RPS27",
groupName = "GMP",
plottype = "QTLplot",
removeoutlier = TRUE)
```
boxplot:
```{r visualizeQTL_sceobject, message=FALSE}
visualizeQTL(
eQTLObject = eqtl1_sce,
SNPid = "1:632647",
Geneid = "RPS27",
groupName = NULL,
plottype = "boxplot",
removeoutlier = TRUE)
```
