原文链接
9、DEG(Differential expressed genes) - 简书

之前DEG分析时在注释celltype的基础上比较condition之间的相同cellty的差异基因。DA分析同样是基于注释celltype，比较不同conditions的相同的celltype的cells数有无显著差异。

• test for significant changes in per-label cell abundance across conditions；
• reveal which cell types are depleted or enriched upon treatment.

1、preparation "count matrics"

• quantify the number of cells assigned to each label (or cluster).
• identify labels that change in abundance among the compartment of injected cells compared to the background.

load("merged.Rdata")
merged
abundances <- table(merged$celltype.mapped, merged$sample) 
#六个sample的34个celltype的cells number
abundances <- unclass(abundances) 
class(abundances)
head(abundances)

1-1

2、

• 接下来的分析流程同样是使用edgeR的分析流程
• 唯一的区别就是the counts are not of reads per gene, but of cells per label

2.1 creat DEGList

extra.info <- colData(merged)[match(colnames(abundances), merged$sample),]
library(edgeR)
y.ab <- DGEList(counts=abundances, samples=extra.info)
y.ab

2-1

2.2 filter out low-abundance labels

keep <- filterByExpr(y.ab, group=y.ab$samples$tomato)
summary(keep)
#去除10个celltype（10行）
y.ab <- y.ab[keep,]

2.3 give design

design <- model.matrix(~factor(pool) + factor(tomato), y.ab$samples)
design

2-2

2.4 DA analysis

#estimateDisp() function to estimate the NB dipersion for each cluster
y.ab <- estimateDisp(y.ab, design, trend="none")
#turn off the trend as we do not have enough points for its stable estimation.
summary(y.ab$common.dispersion)
plotBCV(y.ab, cex=1)
#QL dispersion
fit.ab <- glmQLFit(y.ab, design, robust=TRUE, abundance.trend=FALSE)
summary(fit.ab$var.prior)
summary(fit.ab$df.prior)
plotQLDisp(fit.ab, cex=1)
#test for differences in abundance between td-Tomato-positive and negative samples using glmQLFTest().
res <- glmQLFTest(fit.ab, coef=ncol(design))
summary(decideTests(res))
topTags(res)

• 如下图结果
  （1）extra-embryonic ectoderm is strongly depleted in the injected cells.
  （2）This is consistent with the expectation that cells injected into the blastocyst should not contribute to extra-embryonic tissue.
  （3）The injected cells also contribute more to the mesenchyme, which may also be of interest.

  
  2-4

---

以上是第十五章differential-expression-between-conditions第二部分的简单流程笔记，主要学习了single cell DA分析详见Chapter 14 Multi-sample comparisons
本系列笔记基于OSCA全流程的大致流程梳理，详细原理可参考原文。如有错误，恳请指正！
此外还有刘小泽老师整理的全文翻译笔记，详见目录。