R-DESeq2-SVA

> setwd(工作目录)

> getwd()

[1] "C:/Users/OPT-3070/Desktop"

> library(DESeq2)

> count <- read.csv("counts_MSC.csv", header = T, row.names = 1)#矩阵

> View(count)

> colData <- read.csv("colData.csv", header = T, row.names = 1) #表型，分为两列：condition和indivi

> View(colData)

> dds <- DESeqDataSetFromMatrix(count, colData, design = ~indivi + condition)

> dds

> dds$condition <- relevel(dds$condition, ref = "inj_0")#定义对照

> dds <- dds[rowSums(counts(dds)) >1, ]#去低质量

> dds <- estimateSizeFactors(dds)

> dat <- counts(dds, normalized=TRUE)

> dat <- dat[rowMeans(dat) > 1,]

> head(dat)

> library(sva)

> mod <- model.matrix(~ condition, colData(dds))#构建矩阵

> View(mod)

> mod0 <- model.matrix(~ 1, colData(dds))#构建对照矩阵

> View(mod0)

> svseq <- svaseq(dat, mod, mod0, n.sv = 2)#混杂因素数目n.sv指定为2

> svseq$sv

> dds$SV1 <- svseq$sv[,1]

> dds$SV2 <- svseq$sv[,2]

> design(dds) <- as.formula(paste("~ SV1 + SV2 + ", design))#添加预测出的混杂因子到dds

> dds

> dds <- DESeq(dds)#基于预测出的混杂因素再次分析