🤩 Ragas | 有手就行！分分钟上手的单细胞分析R包！~

生信漫卷

发布于 2024-11-23 09:08:43

16300

代码可运行

文章被收录于专栏：R语言及实用科研软件

运行总次数：0

代码可运行

写在前面

这周末不值班哦，开心的一批！~😀

最近经常去骑车，真的是忘却烦恼的好办法，各位马喽，一起骑车吧🚴🏻！~🫡

做单细胞的小伙伴，刚入门的时候，肯定觉得要学的包好多啊。🫠

今天和大家分享一个R包，Ragas，真的是有手就行，非常好用。👍

熟练的老手可以直接跳过了，我相信你写的code更个性化，更符合你的数据。🦉

用到的包

rm(list = ls())
library(tidyverse)
# BiocManager::install(c("Seurat", "ggplot2", "randomcoloR", "SingleCellExperiment", "Matrix", "future", "dplyr", "reshape2", "scales", "ComplexHeatmap", "grid", "circlize", "gplots", "muscat", "limma", "ggtree", "patchwork", "ggprism", "rstatix", "cowplot", "aplot", "crayon"), update = F)
# devtools::install_github("davidsjoberg/ggsankey", upgrade = "never")
# devtools::install_github("jig4003/Ragas", upgrade = "never")
library(Ragas)
library(ggsci)

示例数据

今天用到的是Ragas 包中的示例数据，Seurat格式。😘

分为健康（cHD）和 SLE （cSLE）儿童的PBMC单细胞测序数据。📈

csle.pbmc.small

构建Pi格式文件

my.pbmc.pi <- CreatePostIntegrationObject(object = csle.pbmc.small)
my.pbmc.pi

聚类可视化

RunDimPlot(object = my.pbmc.pi)

手动注释

RunDimPlot(object = my.pbmc.pi,
           group.by = "cluster.annotation")

FindAllMarkers

运行RunFindAllMarker从Seurat调用FindAllMarkers。😘

my.pbmc.pi <- RunFindAllMarkers(my.pbmc.pi,
                                ident = "cluster.annotation")
my.pbmc.pi

查看一下找到的marker吧。😏

my.pbmc.pi[["markers"]]

Matrix plot

Marker分析完成后，可以使用Matrix plot绘制每个cluster的Top Marker。💪

RunMatrixPlot(my.pbmc.pi,
              markers.key = "Markers|cluster.annotation|AllMarkers|test.use=wilcox", 
              column.anno.name.rot = 45, 
              heatmap.height = 6)

pseudobulk分析

计算表达频率

在运行pseudobulk分析之前，我们需要先进行表达频率分析。🧐

低表达基因将在最终的pseudobulk分析中被过滤掉。😘

my.pbmc.pi <- CalculateExpFreqs(my.pbmc.pi, 
                                ident = "cluster.annotation",
                                verbose = FALSE)

my.pbmc.pi

开始pseudobulk分析

my.pbmc.pi <- RunPseudobulkAnalysis(object = my.pbmc.pi,
                                    ident.var = "cluster.annotation",
                                    group.var = "Groups",
                                    sample.var = "Names",
                                    group.1 = "cSLE",
                                    group.2 = "cHD")
my.pbmc.pi

可视化pseudobulk分析结果

RunDSDotPlot(object = my.pbmc.pi,
             exp.freq.key = "ExpFreq|cluster.annotation|cutoff=0",
             ds.key = "DS|cluster.annotation|edgeR|group=Groups;sample=Names;gp1=cSLE;gp2=cHD;contrast=cSLE-cHD", 
             p.filter = 0.05,
             FC.filter = 4,
             to.adjust = TRUE
                    )

比例分析

按cluster可视化

RunProportionPlot(csle.pbmc.small, ## input can be a Seurat object
                  ident = "cluster.annotation",
                  group.by = "Groups",
                  method = "pooled",
                  pooled.prop.by = "cluster", ## default
                  axis.text.angle = 45,
                  axis.text.size = 8,
                  return.value = 'ggplot'
                  )

堆叠条形图！~📊

my.pbmc.pi <- RunProportionPlot(my.pbmc.pi, ## if input is a Pi object, an updated Pi object will be the returned by default
                                ident = "Groups",
                                group.by = "cluster.annotation",
                                method = "pooled",
                                pooled.prop.by = "cluster", ## default
                                axis.text.size = 8)

## A new data field called "cell.prop" will be added to the Pi object.
my.pbmc.pi

按group可视化

my.pbmc.pi <- RunProportionPlot(my.pbmc.pi,
                                ident = "cluster.annotation",
                                group.by = "Groups",
                                method = "pooled",
                                pooled.prop.by = "group",
                                axis.text.size = 8
                                )
my.pbmc.pi

分组boxplot

my.pbmc.pi <- RunProportionPlot(my.pbmc.pi,
                  ident = "cluster.annotation",
                  group.by = "Groups",
                  method = "unpooled",
                  unpool.by = "Names",
                  unpool.ncol = 3,
                  title.text.size = 6)

分组barplot

my.pbmc.pi <- RunProportionPlot(my.pbmc.pi,
                                ident = "cluster.annotation",
                                group.by = "Groups",
                                method = "unpooled",
                                unpool.by = "Names",
                                unpool.plot.type = "barplot",
                                unpool.ncol = 3,
                                title.text.size = 6)

dotplot可视化features

my.list <- list(Monocytes = c("CD14","FCGR3A"),
                B = "MS4A1",
                T = c("CCR7", "CD8A"),
                NK = "NKG7")

RunAnnotatedDotPlot(object = my.pbmc.pi,
                    annotations = my.list,
                    annotation.cols = c( 'Monocytes'= 'tomato', 'B' = 'seagreen', 'T' = 'steelblue', 'NK' = 'purple'),
                    group.by = "cluster.annotation")

堆积小提琴图可视化features

RunStackedVlnPlot(my.pbmc.pi,
                  ident = "cluster.annotation",
                  features = my.list,
                  color.by = "mean.exp",
                  column.names.rotation = 90)

再加个参数，复杂一点！~🕐

RunStackedVlnPlot(my.pbmc.pi,
                  ident = "cluster.annotation",
                  features = my.list,
                  split.by = "Groups",
                  color.by = "split.var",
                  column.names.rotation = 45)

热图可视化features

features <- list("T cell" = c("CD8A", "CD8B"),
                ISG = c("ISG15","ISG20","IFI44L", "IFI6", "MX1"),
                "Housekeeping" = c("ACTB"))
RunSummarizedHeatmap(object = my.pbmc.pi,
                     ident = "cluster.annotation",
                     features = features,
                     split.by = "Groups", 
                     column.names.rotation = 30,
                     heatmap.width = 20)

再加2个参数，复杂地卷起来！~🧐

RunSummarizedHeatmap(object = my.pbmc.pi,
                     ident = "cluster.annotation",
                     features = features,
                     split.by = "Names", ## sample name
                     additional.metadata = c("Groups", "Gender"),
                     heatmap.width = 60,
                     heatmap.height = 15,
                     show.column.names = FALSE)

表达水平可视化

RunExpressionPlot(object = my.pbmc.pi,
                  feature = "ISG15",
                  ident = "cluster.annotation",
                  group.by = "Names",
                  split.by = "Groups"
                  )+
  theme(
    panel.background = element_blank(), 
    panel.grid.major = element_blank(), 
    panel.grid.minor = element_blank(),  
    panel.border = element_blank(),      
    axis.line = element_line(color = "black") 
  )+
  scale_color_npg()

导出结果

ExportPiData(object = csle.pbmc.pi,
             field = 'ds',
             key = 'DS|subcluster_idents|edgeR|group=Groups;sample=Names;gp1=cSLE;gp2=cHD;contrast=cSLE-cHD',
             file.prefix = 'pbmc.ds' ## A mandatory argument. User should provide an informative prefix (e.g., "csle.pbmc") for exported data
)

最后祝大家早日不卷!~

本文参与腾讯云自媒体同步曝光计划，分享自微信公众号。

原始发表：2024-08-31，如有侵权请联系 cloudcommunity@tencent.com 删除

list