单细胞分析之细胞交互-3：CellChat

---

常用的细胞通讯软件：

• CellphoneDB：是公开的人工校正的，储存受体、配体以及两种相互作用的数据库。此外，还考虑了结构组成，能够描述异构复合物。（配体-受体+多聚体）
• iTALK：通过平均表达量方式，筛选高表达的胚体和受体，根据结果作圈图。（配体-受体）
• CellChat：CellChat将细胞的基因表达数据作为输入，并结合配体受体及其辅助因子的相互作用来模拟细胞间通讯。（配体-受体+多聚体+辅因子）
• NicheNet // NicheNet多样本分析 // 目标基因的配体和靶基因活性预测：通过将相互作用细胞的表达数据与信号和基因调控网络的先验知识相结合来预测相互作用细胞之间的配体-靶标联系的方法。( 配体-受体+信号通路）
  附：NicheNet使用的常见问题汇总

其它细胞互作软件还包括Celltalker，SingleCellSignalR，scTensor和SoptSC（这几个也是基于配体-受体相互作用）

---

CellChat通过综合信号配体、受体及其辅因子基因的表达只与它们之间互作的先验知识对细胞通讯概率建模。在推断出细胞间通讯网络后，CellChat提供了进一步数据探索、分析和可视化的功能。

文章链接：https://www.nature.com/articles/s41467-021-21246-9
视频链接：https://www.youtube.com/watch?v=kc45au1RhNs

CellChat工作流程图：

在进行细胞交互分析的时候，不同分组的样本尽量不要一起进行分析，想要一起分析的时候需要保证不同分组间的细胞种类一致。同一分组的不同生物学重复可以一起分析（很多文献这样做）。

一、单样本分析（可以是同一组的多个生物学重复一起分析）

1. 安装：

devtools::install_github("sqjin/CellChat")

2. 数据准备：

Sys.setenv(RETICULATE_PYTHON="/usr/bin/python3")
# ⚠️要根据自己python3的路径来设置，可以在终端使用which python3来查看 
library(Seurat)
library(SeuratData)
library(tidyverse)
library(CellChat)
library(NMF)
library(ggalluvial)
library(patchwork)
library(ggplot2)
library(svglite)
options(stringsAsFactors = FALSE)

pbmc3k.final <- readRDS("pbmc.rds")
pbmc3k.final@commands$FindClusters  # 也可以看一看作者的其他命令，Seurat是记录其分析过程的。

#以思维导图方式查看pbm3k.final的结构
library(mindr)
(out <- capture.output(str(pbmc3k.final)))
out2 <- paste(out, collapse="\n")

mm(gsub("\\.\\.@","# ",gsub("\\.\\. ","#",out2)),type ="text",root= "Seurat")

3. 创建一个Cell Chat对象

从Seurat对象直接创建：
⚠️：构建Cell Chat对象时，输入的是log后的数据。

cellchat <- createCellChat(object=pbmc3k.final,group.by = "cell_type")
#cellchat <- createCellChat(pbmc3k.final@assays$RNA@data, meta = [email protected], group.by = "cell_type")
cellchat
summary(cellchat)
str(cellchat)
levels(cellchat@idents)
#cellchat <- setIdent(cellchat, ident.use = "cell_type")
groupSize <- as.numeric(table(cellchat@idents))  
#查看每个cluster有多少个细胞，后面画图的时候需要用到这个值
groupSize
# [1] 711 480 472 344 279 162 144  32  14

4. 导入配体受体数据库

CellChatDB <- CellChatDB.human
#导入小鼠是CellChatDB <- CellChatDB.mouse
str(CellChatDB) #查看数据库信息
#包含interaction、complex、cofactor和geneInfo这4个dataframe
colnames(CellChatDB$interaction) 
CellChatDB$interaction[1:4,1:4]
head(CellChatDB$cofactor)
head(CellChatDB$complex)
head(CellChatDB$geneInfo)
#dev.new() #下一步不出图的时候运行
showDatabaseCategory(CellChatDB)

在CellChat中，我们还可以先择特定的信息描述细胞间的相互作用，可以理解为从特定的侧面来刻画细胞间相互作用，比用一个大的配体库又精细了许多。

unique(CellChatDB$interaction$annotation)#查看可以选择的侧面，也就是上图左中的三种
#选择"Secreted Signaling"进行后续细胞互作分析
CellChatDB.use <- subsetDB(CellChatDB, search = "Secreted Signaling") 
# use all CellChatDB for cell-cell communication analysis
# CellChatDB.use <- CellChatDB # simply use the default CellChatDB
# set the used database in the object
cellchat@DB <- CellChatDB.use # set the used database in the object

5. 预处理

对表达数据进行预处理，用于细胞间的通信分析。首先在一个细胞组中识别过表达的配体或受体，然后将基因表达数据投射到蛋白-蛋白相互作用(PPI)网络上。如果配体或受体过表达，则识别过表达配体和受体之间的相互作用。

## 在矩阵的所有的基因中提取signaling gene，结果保存在data.signaling(13714个基因，过滤完只有270个）
cellchat <- subsetData(cellchat)
future::plan("multiprocess", workers = 4)
#相当于Seurat的FindMarkers，找每个细胞群中高表达的配体受体
cellchat <- identifyOverExpressedGenes(cellchat)
cellchat <- identifyOverExpressedInteractions(cellchat) #Identify over-expressed ligand-receptor interactions (pairs) within the used CellChatDB
#上一步运行的结果储存在cellchat@LR$LRsig
cellchat <- projectData(cellchat, PPI.human) 
#找到配体受体关系后，projectData将配体受体对的表达值投射到PPI上，来对@data.signaling中的表达值进行校正。结果保存在@data.project

6. 推断细胞通讯网络

通过为每个相互作用分配一个概率值并进行置换检验来推断生物意义上的细胞-细胞通信。

• 推断配体-受体水平细胞通讯网络（结果储存在@net下面，有一个概率值和对应的pval）
  ⚠️这一步也是CellChat比CellPhoneDB多的一步

通过计算与每个信号通路相关的所有配体-受体相互作用的通信概率来推断信号通路水平上的通信概率。

#根据表达值推测细胞互作的概率（cellphonedb是用平均表达值代表互作强度）。
cellchat <- computeCommunProb(cellchat, raw.use = FALSE, population.size = TRUE) #如果不想用上一步PPI矫正的结果，raw.use = TRUE即可。
# Filter out the cell-cell communication if there are only few number of cells in certain cell groups
cellchat <- filterCommunication(cellchat, min.cells = 10)
df.net <- subsetCommunication(cellchat)
write.csv(df.net, "net_lr.csv")

net_lr（配体-受体水平细胞通讯网络）

• 推断信号通路水平的细胞通讯网络（结果储存在@netP下面，有一个概率值和对应的pval）

我们可以通过计算链路的数量或汇总通信概率来计算细胞间的聚合通信网络。

cellchat <- computeCommunProbPathway(cellchat)
df.netp <- subsetCommunication(cellchat, slot.name = "netP")
write.csv(df.netp, "net_pathway.csv")

net_pathway（信号通路水平的细胞通讯网络）

至此，统计分析部分已经完成。

7. 细胞互作关系展示

7.1 所有细胞群总体观：细胞互作数量与强度统计分析：

#统计细胞和细胞之间通信的数量（有多少个配体-受体对）和强度（概率）
cellchat <- aggregateNet(cellchat)
#计算每种细胞各有多少个
groupSize <- as.numeric(table(cellchat@idents))
par(mfrow = c(1,2), xpd=TRUE)
netVisual_circle(cellchat@net$count, vertex.weight = groupSize, weight.scale = T, 
                     label.edge= F, title.name = "Number of interactions")
netVisual_circle(cellchat@net$weight, vertex.weight = groupSize, weight.scale = T, 
                     label.edge= F, title.name = "Interaction weights/strength")
# save as TIL/net_number_strength.pdf

左图：外周各种颜色圆圈的大小表示细胞的数量，圈越大，细胞数越多。发出箭头的细胞表达配体，箭头指向的细胞表达受体。配体-受体对越多，线越粗。右图：互作的概率/强度值（强度就是概率值相加）

检查每种细胞发出的信号

mat <- cellchat@net$count
par(mfrow = c(3,3), xpd=TRUE)
for (i in 1:nrow(mat)) {
  # i = 1
  mat2 <- matrix(0, nrow = nrow(mat), ncol = ncol(mat), dimnames = dimnames(mat))
  mat2[i, ] <- mat[i, ]
  netVisual_circle(mat2, vertex.weight = groupSize, weight.scale = T, arrow.width = 0.2,
                    arrow.size = 0.1, edge.weight.max = max(mat), title.name = rownames(mat)[i])
}
# save as TIL/net_number_individual.pdf

## 运行上述代码出现报错 Error in plot.new() : figure margins too large
# par("mar")
## [1] 5.1 4.1 4.1 2.1
# par(mar=c(1,1,1,1))
# 重新运行上面的代码

每个细胞如何跟别的细胞互作（number of interaction图）

mat <- cellchat@net$weight
par(mfrow = c(3,3), xpd=T)
for (i in 1:nrow(mat)) {
  mat2 <- matrix(0, nrow = nrow(mat), ncol = ncol(mat), dimnames = dimnames(mat))
  mat2[i, ] <- mat[i, ]
  netVisual_circle(mat2, vertex.weight = groupSize, weight.scale = T, arrow.width = 0.2,
                   arrow.size = 0.1, edge.weight.max = max(mat), title.name = rownames(mat)[i])
}
# save as TIL/net_strength_individual.pdf

每个细胞如何跟别的细胞互作（互作的强度/概率图）

mat <- cellchat@net$weight
par(mfrow = c(2,3), xpd=T)
for (i in 1:nrow(mat)) {
      mat2 <- matrix(0, nrow = nrow(mat), ncol = ncol(mat), dimnames = dimnames(mat))
      mat2[i, ] <- mat[i, ]
      netVisual_circle(mat2, vertex.weight = groupSize, weight.scale = T, arrow.width = 0.2,
                       arrow.size = 0.1, edge.weight.max = max(mat), title.name = rownames(mat)[i])
    }
# save as TIL/net_strength_individual.pdf

7.2 单个信号通路或配体-受体介导的细胞互作可视化（层次图、网络图、和弦图、热图）

⚠️注：层次图、网络图、和弦图、热图只是不同的展示方法，展示的内容和代表的意思一模一样
比如在前面的功能富集分析或case control的比较中找到了一些信号通路差异，就可以进一步在细胞水平上验证。

cellchat@netP$pathways  #查看都有哪些信号通路
# [1] "TGFb"       "NRG"        "PDGF"       "CCL"        "CXCL"       "MIF"        "IL2"        "IL6"       
# [9] "IL10"       "IL1"        "CSF"        "IL16"       "IFN-II"     "LT"         "LIGHT"      "FASLG"     
# [17] "TRAIL"      "BAFF"       "CD40"       "VISFATIN"   "COMPLEMENT" "PARs"       "FLT3"       "ANNEXIN"   
# [25] "GAS"        "GRN"        "GALECTIN"   "BTLA"       "BAG"     
# 选择其中一个信号通路，比如说TGFb
pathways.show <- c("TGFb")  

层次图（Hierarchy plot）

levels(cellchat@idents)    # show all celltype
# [1] "Naive CD4 T"  "Memory CD4 T" "CD14+ Mono"   "B"            "CD8 T"       
# [6] "FCGR3A+ Mono" "NK"           "DC"           "Platelet"    
vertex.receiver = c(1,2,4,6) # define a numeric vector （淋系细胞）giving the index of the celltype as targets
#par(mar=c(5.1,4.1,4.1,2.1))
netVisual_aggregate(cellchat, signaling = pathways.show,  vertex.receiver = vertex.receiver)
# save as TIL/CXCL_hierarchy.pdf

在层次图中，实体圆和空心圆分别表示源和目标。圆的大小与每个细胞组的细胞数成比例。线越粗，互作信号越强。左图中间的target是我们选定的靶细胞。右图是选中的靶细胞之外的另外一组放在中间看互作。

网络图（Circle plot）

par(mfrow=c(1,1))
netVisual_aggregate(cellchat, signaling = pathways.show, layout = "circle")
# save as TIL/CXCL_circle.pdf

和弦图（Chord diagram）

par(mfrow=c(1,1))
netVisual_aggregate(cellchat, signaling = pathways.show, layout = "chord")
# save as TIL/CXCL_chord.pdf

热图（Heatmap）

par(mfrow=c(1,1))
netVisual_heatmap(cellchat, signaling = pathways.show, color.heatmap = "Reds")
# save as TIL/CXCL_heatmap.pdf

纵轴是发出信号的细胞，横轴是接收信号的细胞，热图颜色深浅代表信号强度。上侧和右侧的柱子是纵轴和横轴强度的累积

7.3 配体-受体层级的可视化（计算各个ligand-receptor pair对信号通路的贡献）

#计算配体受体对选定信号通路的贡献值（在这里就是查看哪条信号通路对TGFb贡献最大）
netAnalysis_contribution(cellchat, signaling = pathways.show)
pairLR.TGFb <- extractEnrichedLR(cellchat, signaling = pathways.show, geneLR.return = FALSE) #提取对TGFb有贡献的所有配体受体 
# save as TIL/CXCL_LR_contribution.pdf

层次图（ Hierarchy plot）

#提取对这个通路贡献最大的配体受体对来展示（也可以选择其他的配体受体对）
LR.show <- pairLR.TGFb[1,] 
vertex.receiver = c(1,2,4,6) # a numeric vector
netVisual_individual(cellchat, signaling = pathways.show,  pairLR.use = LR.show, vertex.receiver = vertex.receiver)
# save as TIL/CXCL_hierarchy2.pdf

网络图（Circle plot）

netVisual_individual(cellchat, signaling = pathways.show, pairLR.use = LR.show, layout = "circle")
# save as TIL/CXCL_circle2.pdf

和弦图（Chord diagram）

netVisual_individual(cellchat, signaling = pathways.show, pairLR.use = LR.show, layout = "chord")
# save as TIL/CXCL_chord2.pdf

Error: not enough space for cells at track index '1'.

7.4 自动（批量）保存每个信号通路的互作结果

# Access all the signaling pathways showing significant communications将所有信号通路找出来
pathways.show.all <- cellchat@netP$pathways
# check the order of cell identity to set suitable vertex.receiver
levels(cellchat@idents)
vertex.receiver = c(1,2,4,6) #不画层次图就不需要这一步
dir.create("all_pathways_com_circle") #创建文件夹保存批量画图结果
setwd("all_pathways_com_circle")
for (i in 1:length(pathways.show.all)) {
      # Visualize communication network associated with both signaling pathway and individual L-R pairs
      netVisual(cellchat, signaling = pathways.show.all[i], out.format = c("pdf"),
                vertex.receiver = vertex.receiver, layout = "circle") #绘制网络图
      # Compute and visualize the contribution of each ligand-receptor pair to the overall signaling pathway
      gg <- netAnalysis_contribution(cellchat, signaling = pathways.show.all[i])
      ggsave(filename=paste0(pathways.show.all[i], "_L-R_contribution.pdf"), 
             plot=gg, width = 5, height = 2.5, units = 'in', dpi = 300)
    }
    setwd("../")

7.5多个配体-受体介导的细胞互作关系可视化

气泡图（全部配体受体）

levels(cellchat@idents)
# show all the significant interactions (L-R pairs)
#需要指定受体细胞和配体细胞
p = netVisual_bubble(cellchat, sources.use = c(3,5,7,8,9), 
                     targets.use = c(1,2,4,6), remove.isolate = FALSE)
ggsave("Mye_Lymph_bubble.pdf", p, width = 8, height = 12) #髓系对淋巴的调节
# save as TIL/Mye_Lymph_bubble.pdf

展现细胞之间配体受体关系最直观的图

气泡图（指定信号通路或配体-受体）

#比如制定CCL和CXCL这两个信号通路
netVisual_bubble(cellchat, sources.use = c(3,5,7,8,9), targets.use = c(1,2,4,6), 
                 signaling = c("CCL","CXCL"), remove.isolate = FALSE)

气泡图（指定信号通路或配体-受体并指定细胞）

# show all the significant interactions (L-R pairs) based on user's input
pairLR.use <- extractEnrichedLR(cellchat, signaling = c("CCL","CXCL","TGFb"))
netVisual_bubble(cellchat, sources.use = c(3,6,8), targets.use = c(1,4,5), 
                 pairLR.use = pairLR.use, remove.isolate = TRUE)

CCL，CXCL和TGFb信号通路参与单核细胞和树突状细胞对T细胞的调控作用情况

参与某条信号通路（如TGFb）的所有基因在细胞群中的表达情况展示（小提琴图和气泡图）

## Plot the signaling gene expression distribution
p = plotGeneExpression(cellchat, signaling = "TGFb")
ggsave("TGFb_GeneExpression_vln.pdf", p, width = 8, height = 8)
p = plotGeneExpression(cellchat, signaling = "TGFb", type = "dot")
ggsave("TGFb_GeneExpression_dot.pdf", p, width = 8, height = 6)

8. 通讯网络系统分析（可信度有待考量）

通讯网络系统分析使用了三种算法：社会网络分析、NMF分析和流行学习与分类
⚠️：不同的算法算出来的结果可能会相互矛盾，需要结合生物学知识加以判断

8.1 社会网络分析（通讯网络中的角色识别）

计算网络中心性权重

cellchat <- netAnalysis_computeCentrality(cellchat, slot.name = "netP")

通过计算每个细胞群的网络中心性指标，识别每类细胞在信号通路中的角色/作用C（发送者、接收者、调解者和影响者）

netAnalysis_signalingRole_network(cellchat, signaling = pathways.show, 
                                   width = 15, height = 6, font.size = 10)
# # save as TIL/SNA_CXCL_signalingRole.pdf

识别细胞的信号流模式

ht1 <- netAnalysis_signalingRole_heatmap(cellchat, pattern = "outgoing", font.size = 5)
ht2 <- netAnalysis_signalingRole_heatmap(cellchat, pattern = "incoming", font.size = 5)
ht1 + ht2
# save as TIL/SNA_SignalingPattern.pdf

上图横轴是细胞类型，纵轴是pathway。左图是各个细胞类型中各个通路发出信号的强度，由图是各个细胞类型中各个通路接受信号的强度（上面和右侧的条柱和热图不对应？）

8.2 非负矩阵分解（NMF）识别细胞的通讯模式（这里是一个比较标准的NMF的应用方式）

• 信号输出细胞的模式识别

#计算分解成几个因子(pattern)比较合适（这一步运行比较慢 。在使用NMF对细胞进行亚群细分时，如果不测试的话，最好选择比细胞类型多一点的值）
selectK(cellchat, pattern = "outgoing")
# save as TIL/NMF_outgoing_selectK.pdf

热图

nPatterns = 2 # 挑选曲线中第一个出现下降的点（从2就开始下降了）
cellchat <- identifyCommunicationPatterns(cellchat, pattern = "outgoing", k = nPatterns, 
                                          width = 5, height = 9, font.size = 6)
# save as TIL/NMF_outgoing_comPattern_heatmap.pdf

按selectK算出来的pattern值分为了2个大的pattern，左图纵轴是细胞类型，算法自动将DC、Memory CD4T、FCGR3A+ Mono和Platelet分为了一个pattern（和Pattern 2对应），剩下的五种细胞分为了一个pattern（和Pattern 1对应）。右图纵轴是信号通路，也是被分为了两个大的pattern，上面一个部分是在Pattern 1中比较活跃的通路，下面一个部分是在Pattern 2中比较活跃的通路

冲击图/河流图（river plot）

netAnalysis_river(cellchat, pattern = "outgoing")
# save as TIL/NMF_outgoing_comPattern_river.pdf

气泡图

netAnalysis_dot(cellchat, pattern = "outgoing")
# save as TIL/NMF_outgoing_comPattern_dotplot.pdf

• 信号输入细胞的模式识别

selectK(cellchat, pattern = "incoming") 
# save as TIL/NMF_incoming_selectK.pdf

热图

nPatterns = 2
cellchat <- identifyCommunicationPatterns(cellchat, pattern = "incoming", k = nPatterns, 
                                          width = 5, height = 9, font.size = 6)
# save as TIL/NMF_incoming_comPattern_heatmap.pdf

冲击图

netAnalysis_river(cellchat, pattern = "incoming")
# save as TIL/NMF_incoming_comPattern_river.pdf

气泡图

netAnalysis_dot(cellchat, pattern = "incoming")
# save as TIL/NMF_incoming_comPattern_dotplot.pdf

8.3 信号网络的流行学习与分类

把共同起作用的信号通路归纳在一起，分为基于功能的归纳和基于拓扑结构的归纳

• 基于功能相似性的流行学习与分类

cellchat <- computeNetSimilarity(cellchat, type = "functional")
cellchat <- netEmbedding(cellchat, type = "functional")
cellchat <- netClustering(cellchat, type = "functional")
#Error in do_one(nmeth) : NA/NaN/Inf in foreign function call (arg 1)
p = netVisual_embedding(cellchat, type = "functional", label.size = 3.5)
ggsave("Manifold_functional_cluster.pdf", p, width = 8, height = 6)
#netVisual_embeddingZoomIn(cellchat, type = "functional", nCol = 2)

• 基于拓扑相似性的流行学习与分类

cellchat <- computeNetSimilarity(cellchat, type = "structural")
cellchat <- netEmbedding(cellchat, type = "structural")
cellchat <- netClustering(cellchat, type = "structural")
p = netVisual_embedding(cellchat, type = "structural", label.size = 3.5)
ggsave("Manifold_structural_cluster.pdf", p, width = 8, height = 6)

## The end
saveRDS(cellchat, file = "cellchat.rds")

二、不同分组之间的配对分析

⚠️：配对分析必须保证细胞类型是一样的，才可以进行配对。如果 两个样本的细胞类型不一样又想进行配对分析时，可以用subset把两个样本的细胞类型取成一致的。

1. 数据准备，分别创建CellChat对象

Sys.setenv(RETICULATE_PYTHON="/usr/bin/python3")
library(Seurat)
library(tidyverse)
library(CellChat)
library(NMF)
library(ggalluvial)
library(patchwork)
rm(list = ls())
options(stringsAsFactors = FALSE)

## 创建cellchat对象
### 提取数据子集
scRNA <- readRDS("~/project/Integrate/scRNA.classified.rds")
scRNA$celltype <- scRNA$SingleR
table(scRNA$celltype)
Idents(scRNA) <- 'celltype'
scRNA <- subset(scRNA, idents = c('B cells','CD4+ T cells','CD8+ T cells','Dendritic cells','Monocytes','NK cells'))
scRNA$celltype <- as.factor(as.character(scRNA$celltype))
table(scRNA$orig.ident)
Idents(scRNA) <- 'orig.ident'
sco.til <- subset(scRNA, idents = c('HNC01TIL', 'HNC10TIL', 'HNC20TIL'))
sco.pbmc <- subset(scRNA, idents = c('HNC01PBMC', 'HNC10PBMC', 'HNC20PBMC'))

### 创建cellchat对象
cco.til <- createCellChat(sco.til@assays$RNA@data, meta = [email protected], group.by = "celltype")
cco.pbmc <- createCellChat(sco.pbmc@assays$RNA@data, meta = [email protected], group.by = "celltype")
save(cco.til, cco.pbmc, file = "cco.rda")

2. 细胞通讯网络分析

• 2.1 数据准备和路径切换

dir.create("./Compare")
setwd("./Compare")
# load("../cco.rda")
# cco.pbmc <- setIdent(cco.pbmc, ident.use = "celltype")
# cco.til <- setIdent(cco.til, ident.use = "celltype")

• 2.2 分析样本cco.pbmc的细胞通讯网络
  ⚠️：cellchat不太稳定，identifyOverExpressedGenes常出错（不出现进度条就是出错了），重启Rstudio后再运算。

cellchat <- cco.pbmc
cellchat@DB <- CellChatDB.human
cellchat <- subsetData(cellchat)
cellchat <- identifyOverExpressedGenes(cellchat)
cellchat <- identifyOverExpressedInteractions(cellchat)
#cellchat <- projectData(cellchat, PPI.human)
cellchat <- computeCommunProb(cellchat, raw.use = TRUE, population.size = TRUE)
#cellchat <- filterCommunication(cellchat, min.cells = 5)
cellchat <- computeCommunProbPathway(cellchat)
cellchat <- aggregateNet(cellchat)
cellchat <- netAnalysis_computeCentrality(cellchat, slot.name = "netP")
#cellchat <- computeNetSimilarity(cellchat, type = "functional")
#cellchat <- netEmbedding(cellchat, type = "functional")
#cellchat <- netClustering(cellchat, type = "functional")
#cellchat <- computeNetSimilarity(cellchat, type = "structural")
#cellchat <- netEmbedding(cellchat, type = "structural")
#cellchat <- netClustering(cellchat, type = "structural")
cco.pbmc <- cellchat
saveRDS(cco.pbmc, "cco.pbmc.rds")

• 2.3 分析样本cco.til的细胞通讯网络

cellchat <- cco.til
cellchat@DB <- CellChatDB.human
cellchat <- subsetData(cellchat)
cellchat <- identifyOverExpressedGenes(cellchat)
cellchat <- identifyOverExpressedInteractions(cellchat)
#cellchat <- projectData(cellchat, PPI.human)
cellchat <- computeCommunProb(cellchat, raw.use = TRUE, population.size = TRUE)
#cellchat <- filterCommunication(cellchat, min.cells = 5)
cellchat <- computeCommunProbPathway(cellchat)
cellchat <- aggregateNet(cellchat)
cellchat <- netAnalysis_computeCentrality(cellchat, slot.name = "netP")
#cellchat <- computeNetSimilarity(cellchat, type = "functional")
#cellchat <- netEmbedding(cellchat, type = "functional")
#cellchat <- netClustering(cellchat, type = "functional")
#cellchat <- computeNetSimilarity(cellchat, type = "structural")
#cellchat <- netEmbedding(cellchat, type = "structural")
#cellchat <- netClustering(cellchat, type = "structural")
cco.til <- cellchat
saveRDS(cco.til, "cco.til.rds")

• 2.4 合并cellchat对象

cco.list <- list(pbmc=cco.pbmc, til=cco.til)
cellchat <- mergeCellChat(cco.list, add.names = names(cco.list), cell.prefix = TRUE)

3. 可视化

3.1 所有细胞群总体观：通讯数量与强度对比

gg1 <- compareInteractions(cellchat, show.legend = F, group = c(1,2), measure = "count")
gg2 <- compareInteractions(cellchat, show.legend = F, group = c(1,2), measure = "weight")
p <- gg1 + gg2
ggsave("Overview_number_strength.pdf", p, width = 6, height = 4)

左图展示通讯数量之间的差异，右图展示通讯强度之间的差异。本例中信号通路强度weight值过低，导致显示时均为0（实际上有数值的，只是过小，显示为0）

数量与强度差异网络图

par(mfrow = c(1,2))
netVisual_diffInteraction(cellchat, weight.scale = T)
netVisual_diffInteraction(cellchat, weight.scale = T, measure = "weight")
# save as Diff_number_strength_net.pdf

红色是case相对于control上调的，蓝色是下调的。

数量与强度差异热图

par(mfrow = c(1,1))
h1 <- netVisual_heatmap(cellchat)
h2 <- netVisual_heatmap(cellchat, measure = "weight")
h1+h2
# save as Diff_number_strength_heatmap.pdf

case和control对比，红色是上调，蓝色是下调。

细胞互作数量对比网络图

par(mfrow = c(1,2))
weight.max <- getMaxWeight(cco.list, attribute = c("idents","count"))
for (i in 1:length(cco.list)) {
  netVisual_circle(cco.list[[i]]@net$count, weight.scale = T, label.edge= F, 
                   edge.weight.max = weight.max[2], edge.width.max = 12, 
                   title.name = paste0("Number of interactions - ", names(cco.list)[i]))
}
# save as Counts_Compare_net.pdf

左图是control，右图是case，可以直接对比数量变化。

3.2 指定细胞互作数量对比网络图

par(mfrow = c(1,2))
s.cell <- c("CD4+ T cells", "CD8+ T cells", "Monocytes")
count1 <- cco.list[[1]]@net$count[s.cell, s.cell]
count2 <- cco.list[[2]]@net$count[s.cell, s.cell]
weight.max <- max(max(count1), max(count2))
netVisual_circle(count1, weight.scale = T, label.edge= T, edge.weight.max = weight.max, edge.width.max = 12, 
                 title.name = paste0("Number of interactions-", names(cco.list)[1]))
netVisual_circle(count2, weight.scale = T, label.edge= T, edge.weight.max = weight.max, edge.width.max = 12, 
                 title.name = paste0("Number of interactions-", names(cco.list)[2]))
# save as Counts_Compare_select.pdf 10*6.5

3.3 保守和特异性信号通路的识别与可视化

## 通路信号强度对比分析
gg1 <- rankNet(cellchat, mode = "comparison", stacked = T, do.stat = TRUE)
gg2 <- rankNet(cellchat, mode = "comparison", stacked = F, do.stat = TRUE)
p <- gg1 + gg2
ggsave("Compare_pathway_strengh.pdf", p, width = 10, height = 6)

左图最下面5个信号通路是case组独有的

3.4 流行学习识别差异信号通路

这里function的图出不来，只有structural的图可以出来

cellchat <- computeNetSimilarityPairwise(cellchat, type = "functional")
cellchat <- netEmbedding(cellchat, type = "functional")
cellchat <- netClustering(cellchat, type = "functional")
#netVisual_embeddingPairwise(cellchat, type = "functional", label.size = 3.5)
#netVisual_embeddingPairwiseZoomIn(cellchat, type = "functional", nCol = 2)
cellchat <- computeNetSimilarityPairwise(cellchat, type = "structural")
cellchat <- netEmbedding(cellchat, type = "structural")
cellchat <- netClustering(cellchat, type = "structural")
#netVisual_embeddingPairwise(cellchat, type = "structural", label.size = 3.5)
#netVisual_embeddingPairwiseZoomIn(cellchat, type = "structural", nCol = 2)
p <- rankSimilarity(cellchat, type = "structural") + ggtitle("Structural similarity of pathway")
ggsave("Pathway_Similarity.pdf", p, width = 8, height = 5)
    
saveRDS(cellchat, "cellchat.rds")

case和control之间信号通路差异相差程度排行，在这张图中，ICAM相差最大，其次是SELPLG。⚠️与上面那张图的结果不相符♀️，个人更倾向于相信上一张图的结果

3.5 细胞信号模式对比

library(ComplexHeatmap)

总体信号模式对比

pathway.union <- union(cco.list[[1]]@netP$pathways, cco.list[[2]]@netP$pathways)
ht1 = netAnalysis_signalingRole_heatmap(cco.list[[1]], pattern = "all", signaling = pathway.union, 
                                        title = names(cco.list)[1], width = 8, height = 10)
ht2 = netAnalysis_signalingRole_heatmap(cco.list[[2]], pattern = "all", signaling = pathway.union,
                                        title = names(cco.list)[2], width = 8, height = 10)
draw(ht1 + ht2, ht_gap = unit(0.5, "cm"))
# save as Compare_signal_pattern_all.pdf  10*6

左图下面几个信号通路在pbmc组中没有，和3.3中的图相符

输出信号模式对比

pathway.union <- union(cco.list[[1]]@netP$pathways, cco.list[[2]]@netP$pathways)
ht1 = netAnalysis_signalingRole_heatmap(cco.list[[1]], pattern = "outgoing", signaling = pathway.union, 
                                        title = names(cco.list)[1], width = 8, height = 10)
ht2 = netAnalysis_signalingRole_heatmap(cco.list[[2]], pattern = "outgoing", signaling = pathway.union,
                                        title = names(cco.list)[2], width = 8, height = 10)
draw(ht1 + ht2, ht_gap = unit(0.5, "cm"))
# save as Compare_signal_pattern_outgoing.pdf  10*6

输入信号模式对比

pathway.union <- union(cco.list[[1]]@netP$pathways, cco.list[[2]]@netP$pathways)
ht1 = netAnalysis_signalingRole_heatmap(cco.list[[1]], pattern = "incoming", signaling = pathway.union, 
                                        title = names(cco.list)[1], width = 8, height = 10)
ht2 = netAnalysis_signalingRole_heatmap(cco.list[[2]], pattern = "incoming", signaling = pathway.union,
                                        title = names(cco.list)[2], width = 8, height = 10)
draw(ht1 + ht2, ht_gap = unit(0.5, "cm"))
# save as Compare_signal_pattern_incoming.pdf  10*6

3.6 特定信号通路的对比

网络图

pathways.show <- c("IL16") 
weight.max <- getMaxWeight(cco.list, slot.name = c("netP"), attribute = pathways.show) 
par(mfrow = c(1,2), xpd=TRUE)
for (i in 1:length(cco.list)) {
  netVisual_aggregate(cco.list[[i]], signaling = pathways.show, layout = "circle", 
                      edge.weight.max = weight.max[1], edge.width.max = 10, 
                      signaling.name = paste(pathways.show, names(cco.list)[i]))
}
# save as Compare_IL16_net.pdf  10*6.5

热图

par(mfrow = c(1,2), xpd=TRUE)
ht <- list()
for (i in 1:length(cco.list)) {
  ht[[i]] <- netVisual_heatmap(cco.list[[i]], signaling = pathways.show, color.heatmap = "Reds",
                               title.name = paste(pathways.show, "signaling ",names(cco.list)[i]))
}
ComplexHeatmap::draw(ht[[1]] + ht[[2]], ht_gap = unit(0.5, "cm"))
# save as Compare_IL16_heatmap.pdf  12*6.5

和弦图

par(mfrow = c(1,2), xpd=TRUE)
for (i in 1:length(cco.list)) {
  netVisual_aggregate(cco.list[[i]], signaling = pathways.show, layout = "chord", pt.title = 3, title.space = 0.05,
                      vertex.label.cex = 0.6, signaling.name = paste(pathways.show, names(cco.list)[i]))
}
# save as Compare_IL16_chord.pdf  10*6.5

3.7 配体-受体对比分析

气泡图展示所有配体受体对的差异

levels(cellchat@idents$joint)
p <- netVisual_bubble(cellchat, sources.use = c(4,5), targets.use = c(1,2,3,6),  comparison = c(1, 2), angle.x = 45)
ggsave("Compare_LR_bubble.pdf", p, width = 12, height = 8)

case和control配对的图，文章中常见。cellphoneDB找到的结果，也可以用这种方式呈现。

气泡图展示上调或下调的配体受体对

p1 <- netVisual_bubble(cellchat, sources.use = c(4,5), targets.use = c(1,2,3,6), comparison = c(1, 2), 
                       max.dataset = 2, title.name = "Increased signaling in TIL", angle.x = 45, remove.isolate = T)
p2 <- netVisual_bubble(cellchat, sources.use = c(4,5), targets.use = c(1,2,3,6), comparison = c(1, 2), 
                       max.dataset = 1, title.name = "Decreased signaling in TIL", angle.x = 45, remove.isolate = T)
pc <- p1 + p2
ggsave("Compare_LR_regulated.pdf", pc, width = 12, height = 5.5)

上调下调分开展示

和弦图

par(mfrow = c(1, 2), xpd=TRUE)
for (i in 1:length(cco.list)) {
  netVisual_chord_gene(cco.list[[i]], sources.use = c(4,5), targets.use = c(1,2,3,6), signaling = "MHC-I", 
                       lab.cex = 0.6, legend.pos.x = 10, legend.pos.y = 20,
                       title.name = paste0("Signaling from Treg - ", names(cco.list)[i]))
}
# save as Compare_LR_chord.pdf  10*6.5

小思考：但组间细胞类型占比变化似乎对CellChat组间分析结果影响较大，尤其是细胞互作强度。
如处理组和对照组相比，巨噬细胞显著减少，单核细胞显著增多的情况下，蛋壳图中巨噬细胞和巨噬细胞的互作强度（注意 不是数量）就会显著下降，单核和单核的互作强度就会显著上升。
由于互作强度是根据表达的受体配体数决定的，那么细胞占比增多，强度就会增加，细胞占比下降，强度就下降？这样似乎不是很科学

---

应用

• 单细胞核测序和空间转录组解析肥厚性心肌病的心脏病理性重构的谱系特异性调节改变