新版 CellphoneDB v5 新增功能：ktplots绘图包尝鲜

生信技能树

发布于 2025-05-17 15:16:01

54300

代码可运行

文章被收录于专栏：生信技能树生信技能树

运行总次数：0

代码可运行

❝前面提到一眨眼CellphoneDB都更新到v5啦，今天就来看看它更新的绘图功能！ ❞

老习惯，CellphoneDB 的官网给到你：https://cellphonedb.readthedocs.io/en/latest/

CellphoneDB 的一路更新历程，依然放出来，你每篇文献都下载下来看了吗看了吗看了吗？

「CellphoneDB v1 (original)」: Single-cell reconstruction of the early maternal-fetal interface in humans. Vento-Tormo R, Efremova M, et al., Nature. 2018
「CellphoneDB v2」: Inferring cell-cell communication from combined expression of multi-subunit receptor-ligand complexes. Efremova M, Vento-Tormo M, Teichmann S, Vento-Tormo R. Nat Protoc. 2020
「CellphoneDB v3」: Mapping the temporal and spatial dynamics of the human endometrium in vivo and in vitro. L Garcia-Alonso, L-François Handfield, K Roberts, K Nikolakopoulou et al. Nature Genetics 2021
「CellphoneDB v4」: Single-cell roadmap of human gonadal development. L Garcia-Alonso, V Lorenzi et al. 2022 Nature
「CellphoneDB v5 (latest)」: CellphoneDB v5: inferring cell-cell communication from single-cell multiomics data.

CellphoneDB 对结果的可视化开发了两个包，一个R包和一个对应的 python版本包：

ktplots (https://www.github.com/zktuong/ktplots/) (R; preferred)
ktplotspy (https://www.github.com/zktuong/ktplotspy/) (python; under development)

先来看看 R版本的 ktplots！

0.软件安装

## 使用西湖大学的 Bioconductor镜像
options(BioC_mirror="https://mirrors.westlake.edu.cn/bioconductor")
options("repos"=c(CRAN="https://mirrors.westlake.edu.cn/CRAN/"))

# 在线安装有点不好安装
# devtools::install_github("zktuong/ktplots", dependencies = TRUE)
# https://github.com/zktuong/ktplots/ 这里下载安装包 ktplots-master.zip
# 本地安装
devtools::install_local("ktplots-master.zip")

# 加载R包
library(ktplots)

示例数据

这个数据来自：https://www.kidneycellatlas.org/，对应的文献于2019年9月27号发表在Science杂志上，文献标题为《Spatiotemporal immune zonation of the human kidney》。这里作者直接放到了R包中，经过了降采样即抽取部分细胞的cellphoneDB结果。

此包绘图需要3个文件：

一个用于 CellPhoneDB 的对象相对应的 SingleCellExperiment（或Seurat；一些函数仅接受前者）对象；
运行完 CellPhoneDB 后的 means.txt 和 pvalues.txt 文件；
如果使用的是从版本>= 3的CellPhoneDB deg_analysis模式的结果，则 pvalues.txt 是 relevant_interactions.txt

这里绘图需要理解一下 CellPhoneDB的结果 means.txt 和 pvalues.txt 每一列的含义，详细解释见：

https://cellphonedb.readthedocs.io/en/latest/RESULTS-DOCUMENTATION.html#output-files

library(SingleCellExperiment)
data(kidneyimmune)

# pvals <- read.delim("pvalues.txt", check.names = FALSE)
# means <- read.delim("means.txt", check.names = FALSE)
# decon = pd.read_csv("deconvoluted.txt", sep="\t")

# I've provided an example datasets
data(cpdb_output_stat) # ran with CellPhoneDB statistical analysis mode
data(cpdb_output_degs) # ran with CellPhoneDB degs analysis mode

绘制细胞通讯结果

1.热图

「用于显示每个细胞类型对之间显著相互作用的数量」

使用这个重新实现的函数可以生成CellPhoneDB的原始热图图示：

# 互作热图
dim(pvals_stat)
pvals_stat[1:5,1:5]
# 绘图
plot_cpdb_heatmap(pvals = pvals_stat, cellheight = 10, cellwidth = 10)

差异分析模式：

# degs_analysis = TRUE
dim( rel_int_degs)
rel_int_degs[1:5,1:5]
plot_cpdb_heatmap(pvals = rel_int_degs, cellheight = 10, cellwidth = 10, degs_analysis = TRUE)

指定细胞亚群:

# 指定细胞亚群
dim(pvals_stat)
pvals_stat[1:5,1:5]
plot_cpdb_heatmap(pvals = pvals_stat, cell_types = c("NK cell", "pDC", "B cell", "CD8T cell"), 
                  cellheight = 10, cellwidth = 10)

2.气泡图

「显示所有选定细胞类型/相互作用对的显著相互作用。较大的点表示细胞类型对（x轴）的相互作用（y轴）是相关的，而颜色表示平均表达。」

plot_cpdb 函数：

可以指定某些特定配体受体，以及细胞亚群之间的通讯情况，下图展示的所有B cell 与其他所有细胞类型的有 "PTPRC", "TNFSF13" 这两个基因的通讯结果：

# 气泡图
kidneyimmune
anno = colData(kidneyimmune)

plot_cpdb(
  scdata = kidneyimmune,
  cell_type1 = "B cell",
  cell_type2 = ".", # this means all cell-types
  celltype_key = "celltype",
  means = means_stat,
  pvals = pvals_stat,
  genes = c("PTPRC", "TNFSF13"),
  title = "interacting interactions!",
)

这个图跟以前的不一样还加了一个红色的外圈来表示这个通讯是不是具有统计学意义呀！

配受体互作的名字可以保持原始数据库中的：设置 keep_id_cp_interaction = TRUE

# 设置 keep_id_cp_interaction = TRUE
plot_cpdb(
    scdata = kidneyimmune,
    cell_type1 = "B cell",
    cell_type2 = ".", # this means all cell-types
    celltype_key = "celltype",
    means = means_stat,
    pvals = pvals_stat,
    genes = c("PTPRC", "TNFSF13"),
    title = "interacting interactions!",
    keep_id_cp_interaction = TRUE,
)

配受体互作名字：

默认绘制所有显著意义的互作结果：

# 不指定配受体
plot_cpdb(
  scdata = kidneyimmune,
  cell_type1 = "B cell",
  cell_type2 = ".",
  celltype_key = "celltype",
  means = means_stat,
  pvals = pvals_stat,
)
ggsave(filename = "B_cells.png", width = 12,height = 10,bg="white")

还可以更改加的红色外圈展示方式，这种方式受到squidpy包的启发，设置 default_style = FALSE

# 设置 default_style = FALSE
plot_cpdb(
    scdata = kidneyimmune,
    cell_type1 = "B cell",
    cell_type2 = ".",
    celltype_key = "celltype",
    means = means_stat,
    pvals = pvals_stat,
    genes = c("PTPRC", "CD40", "CLEC2D"),
    default_style = FALSE
)

还可以对细胞亚群进行分组展示，以及配受体展示某一具体基因相关家族如细胞因子家族chemokines：

设置 splitby_key = "Experiment" 和 gene_family = "chemokines"

# 对细胞亚群进行分组展示
data(cpdb_output) # this is a different dataset where the "Experiment" was appended to the "celltype"

plot_cpdb(
    scdata = kidneyimmune,
    cell_type1 = "B cell",
    cell_type2 = "Neutrophil|MNPc|NK cell",
    celltype_key = "celltype",
    means = means,
    pvals = pvals,
    splitby_key = "Experiment",
    gene_family = "chemokines"
)

还接受指定的配受体表格：

# 提供指定配受体表格绘制
df <- data.frame(set1 = c("CCR6", "CCL20", "CXCL10", "CCR3", "TNFRSF13C"), 
                 set2 = c("CCL5", "CCR4", "PTPRC", "CD40", "CLEC2D"))
df

plot_cpdb(
  scdata = kidneyimmune,
  cell_type1 = "B cell",
  cell_type2 = "CD4T cell",
  celltype_key = "celltype",
  means = means,
  pvals = pvals,
  splitby_key = "Experiment",
  gene_family = c("set1", "set2"),
  custom_gene_family = df,
)

3.玄图展示

「plot_cpdb2 为cellphonedb结果生成circos风格的线/弧/和弦图」

这个函数也是新增的：

### 玄图1
data(cpdb_output2) # legacy reasons

plot_cpdb2(
  scdata = kidneyimmune,
  cell_type1 = "B cell",
  cell_type2 = ".",
  celltype_key = "celltype", # column name where the cell ids are located in the metadata
  means = means2,
  pvals = pvals2,
  deconvoluted = decon2, # new options from here on specific to plot_cpdb2
  desiredInteractions = list( c("CD4T cell", "B cell"), c("B cell", "CD4T cell") ),
  interaction_grouping = interaction_annotation,
  edge_group_colors = c(
    "Activating" = "#e15759",
    "Chemotaxis" = "#59a14f",
    "Inhibitory" = "#4e79a7",
    "Intracellular trafficking" = "#9c755f",
    "DC_development" = "#B07aa1",
    "Unknown" = "#e7e7e7"
  ),
  node_group_colors = c( "CD4T cell" = "red", "B cell" = "blue" ),
)

这里展示了 "CD4T cell", "B cell" 这两种细胞间的具体通讯结果，通讯的连线不同的颜色代表不同的互作情况。

圆圈的颜色表示细胞类型，蓝色为 "B cell"，红色为 "CD4T cell"。

plot_cpdb3

「生成一个灵感来源于CellChat展示数据方式的和弦图！用法与plot_cpdb2相似，但选项减少」

## 与cellchat类似的展示方式
plot_cpdb3(
  scdata = kidneyimmune,
  cell_type1 = "B cell",
  cell_type2 = "CD4T cell|MNPd",
  celltype_key = "celltype", # column name where the cell ids are located in the metadata
  means = means_stat,
  pvals = pvals_stat,
  deconvoluted = decon_stat # new options from here on specific to plot_cpdb3
)

plot_cpdb4

「新功能！展示特定交互的弦图的新方法！用法与plot_cpdb3类似，但需要额外的交互选项」

# new
plot_cpdb4(
  scdata = kidneyimmune,
  interaction = "CLEC2D-KLRB1",
  cell_type1 = "NK",
  cell_type2 = "Mast",
  celltype_key = "celltype",
  means = means_stat,
  pvals = pvals_stat,
  deconvoluted = decon_stat
)

展示的是 "CLEC2D-KLRB1" 这对配受体在不同亚群之间的情况。

其他功能

这里还有一个函数 geneDotPlot 可以绘制基因的表达：

## 基因表达
library(ggplot2)
# Note, this conflicts with tidyr devel version
p <- geneDotPlot(
  scdata = kidneyimmune, # object
  genes = c("CD68", "CD80", "CD86", "CD74", "CD2", "CD5"), # genes to plot
  celltype_key = "celltype", # column name in meta data that holds the cell-cluster ID/assignment
  splitby_key = "Project", # column name in the meta data that you want to split the plotting by. If not provided, it will just plot according to celltype_key
  standard_scale = TRUE ) + 
  theme(strip.text.x = element_text(angle = 0, hjust = 0, size = 7)) + 
  small_guide() + 
  small_legend()
p