🤩 SeuratExtend | 基于Seurat的强大拓展包！~（二）（基础可视化）

生信漫卷

发布于 2025-11-13 18:32:35

写在前面

本期内容：👇

VlnPlot2：一行命令完成点样式/分组拆分/均值&中位线/离群点隐藏/统计检验；
ClusterDistrBar / ClusterDistrPlot：一眼看懂不同样本/条件的群集分布（百分比/绝对数、堆叠/非堆叠、横/纵向可切换）；
WaterfallPlot：用于通路打分或基因集的差异“瀑布图”（支持显著性过滤、Log2/Log10、TopN）；
VolcanoPlot：单细胞友好的火山图（支持 t-score 作为 Y 轴，避免“大样本 p 值挤顶”）。

用到的包

rm(list = ls())
library(Seurat)
library(SeuratExtend)

示例数据

pbmc

小提琴图

基本用法

genes <- c("CD3D","CD14","CD79A")

VlnPlot2(pbmc, features = genes, ncol = 1)

点随机分布！~🫠

VlnPlot2(pbmc, features = genes, violin = FALSE, pt.style = "quasirandom", ncol = 1)

不显示点！~😀

VlnPlot2(pbmc, features = genes, pt = FALSE, ncol = 1)

不显示outlier！~😅

VlnPlot2(pbmc, features = genes, pt = FALSE, hide.outlier = TRUE, ncol = 1)

Outline Style

VlnPlot2(pbmc, features = genes, style = "outline", ncol = 1)

添加平均线和中位数线

lowExprGenes <- c("CCR7", "IL7R", "TCF7")
VlnPlot2(pbmc, 
         features = lowExprGenes, 
         show.mean = TRUE,      # Show mean and median lines
         mean_colors = c("red", "blue"),  # Colors for mean and median
         cols = "light",        # Light color scheme for better visibility
         ncol = 1)

拆分可视化

VlnPlot2(pbmc, features = genes, group.by = "cluster", split.by = "orig.ident")

子集分析

cells <- colnames(pbmc)[pbmc$cluster %in% c("B cell", "Mono CD14", "CD8 T cell")]
VlnPlot2(pbmc, features = genes, group.by = "cluster", cells = cells)

统计分析

wilcox.test！~

VlnPlot2(pbmc, 
         features = genes, 
         group.by = "cluster", 
         cells = cells,
         stat.method = "wilcox.test", 
         hide.ns = TRUE)

t.test！~

VlnPlot2(pbmc, 
         features = genes, 
         group.by = "cluster", 
         cells = cells,
         stat.method = "t.test", 
         comparisons = list(c(1,2), c(1,3)), 
         hide.ns = FALSE)

使用矩阵作为输入文件

这里我们把基因集富集分析的AUCell分数作为输入文件。😘

# Perform gene set analysis
pbmc <- GeneSetAnalysis(pbmc, genesets = hall50$human)
matr <- pbmc@misc$AUCell$genesets

# Plot the first three pathways
VlnPlot2(matr[1:3,], f = pbmc$cluster, ncol = 1)

可视化Cluster分布

基本用法

ClusterDistrBar(origin = pbmc$orig.ident, cluster = pbmc$cluster)

绝对细胞计数

ClusterDistrBar(origin = pbmc$orig.ident, cluster = pbmc$cluster, percent = FALSE)

垂直方向

ClusterDistrBar(origin = pbmc$orig.ident, cluster = pbmc$cluster, flip = FALSE, reverse_order = FALSE)

非堆叠显示

ClusterDistrBar(origin = pbmc$orig.ident, 
                cluster = pbmc$cluster, 
                flip = FALSE, 
                stack = FALSE)

归一化比例

ClusterDistrBar(origin = pbmc$orig.ident, cluster = pbmc$cluster, rev = TRUE, normalize = TRUE)

比较不同分组之间的细胞分布

# Compare cluster distribution between conditions
ClusterDistrPlot(
  origin = pbmc$sample_id,
  cluster = pbmc$cluster,
  condition = pbmc$condition
)

瀑布图

基本用法

pbmc <- GeneSetAnalysis(pbmc, genesets = hall50$human)
matr <- pbmc@misc$AUCell$genesets

WaterfallPlot(matr, f = pbmc$cluster, ident.1 = "Mono CD14", ident.2 = "CD8 T cell")

segment样式

# Create a plot using the new segment style
WaterfallPlot(
  matr,
  f = pbmc$cluster,
  ident.1 = "Mono CD14",
  ident.2 = "CD8 T cell",
  style = "segment",
  color_theme = "D"
)

按显著性过滤

WaterfallPlot(
  matr, 
  f = pbmc$cluster, 
  ident.1 = "Mono CD14", 
  ident.2 = "CD8 T cell", 
  len.threshold = 2)

比较基因表达

genes <- VariableFeatures(pbmc)[1:70]
WaterfallPlot(
  pbmc, 
  group.by = "cluster", 
  features = genes,
  ident.1 = "Mono CD14", 
  ident.2 = "CD8 T cell", 
  length = "logFC")

仅显示差异表达较大的基因！~🥸

WaterfallPlot(
  pbmc, 
  group.by = "cluster", 
  features = genes,
  ident.1 = "Mono CD14", 
  ident.2 = "CD8 T cell", 
  length = "logFC",
  top.n = 20)

使用Log2或Log10展示倍数变化。😀

# Using log base 2 for fold change calculations
WaterfallPlot(
  pbmc, 
  group.by = "cluster", 
  features = genes,
  ident.1 = "Mono CD14", 
  ident.2 = "CD8 T cell", 
  length = "logFC",
  log.base = "2",    # Use log2 instead of natural log
  top.n = 20)

火山图

基本用法

VolcanoPlot(pbmc, 
            ident.1 = "B cell",
            ident.2 = "CD8 T cell")

自定义阈值

VolcanoPlot(
  pbmc,
  ident.1 = "B cell",
  ident.2 = "CD8 T cell",
  x.threshold = 0.5,    # Log fold change threshold
  y.threshold = 2     # -log10(p-value) threshold
)

使用 t-score代替p值

在单细胞分析中使用t-scor代替p值作为y轴`特别有用。🥳

由于单细胞数据集中的细胞数量较多，p 值通常变得极其显著，导致许多基因在使用-log10（p 值）时聚集在 y 轴的顶部。🧐

这使得很难区分在技术上都非常重要的基因。🧬

t-score提供了更好的分布，并有助于创建信息更丰富的可视化效果。🙊

VolcanoPlot(
  pbmc,
  ident.1 = "B cell",
  ident.2 = "CD8 T cell",
  y = "tscore"
)

使用 Log2 或 Log10 倍数变化

VolcanoPlot(
  pbmc,
  ident.1 = "B cell",
  ident.2 = "CD8 T cell",
  log.base = "2"    # Use log2 instead of natural log
)

使用矩阵作为输入文件

# Using the previously calculated pathway enrichment scores
VolcanoPlot(
  matr,
  f = pbmc$cluster,
  ident.1 = "B cell",
  ident.2 = "Mono CD14",
  x.quantile = 0.8,
  y.quantile = 0.8,
  top.n = 5)