哈喽小伙伴们，seurat官网整理了一些常用命令，方便我们调用。一起来看看吧！
Standard Seurat workflow（Seurat标准工作流程）
Seurat Object Data Access（Seurat对象数据访问）
Subsetting and merging（子集和合并）
Pseudobulk analysis（Pseudobulk分析）
Visualization in Seurat（Seurat的可视化）
Multi-Assay Features（多检测特性）
Additional resources（其他资源）
pbmc <- NormalizeData(object = pbmc)pbmc <- FindVariableFeatures(object = pbmc)pbmc <- ScaleData(object = pbmc)pbmc <- RunPCA(object = pbmc)pbmc <- FindNeighbors(object = pbmc, dims = 1:30)pbmc <- FindClusters(object = pbmc)pbmc <- RunUMAP(object = pbmc, dims = 1:30)DimPlot(object = pbmc, reduction = "umap")
SCtransform 版本
pbmc <- SCTransform(object = pbmc)pbmc <- RunPCA(object = pbmc)pbmc <- FindNeighbors(object = pbmc, dims = 1:30)pbmc <- FindClusters(object = pbmc)pbmc <- RunUMAP(object = pbmc, dims = 1:30)DimPlot(object = pbmc, reduction = "umap")
#可以使用%>%将多个命令链接在一起pbmc <- SCTransform(pbmc) %>%RunPCA() %>%FindNeighbors(dims = 1:30) %>%FindClusters() %>%RunUMAP(dims = 1:30)
1. 细胞、基因和layer名称
#获取细胞和基因名以及总数的多种方法colnames(pbmc)Cells(pbmc)#基因名称Features(pbmc)rownames(pbmc)#细胞/基因的数量num_cells <- ncol(pbmc)num_features <- nrow(pbmc)#对象层列表Layers(pbmc)#使用多模态对象列表分析Assays(cbmc)#检测特异性特征(基因/ADT)Features(cbmc[["RNA"]])Features(cbmc[["ADT"]])#可变基因名称VariableFeatures(pbmc)
#设置可变基因VariableFeatures(cbmc) <- var.gene.names#设置一个特定的assayVariableFeatures(cbmc[["ADT"]]) <- var.gene.names
2. 确定分类标签
#设置和检索细胞标识#将标识类设置为metadata中的现有列Idents(object = pbmc) <- "seurat_annotations"#查看细胞标识，获取汇总表Idents(pbmc)table(Idents(pbmc))#将所有细胞的标识设置为CD4 T细胞Idents(pbmc) <- "CD4 T cells"#为选定的细胞组设置pbmc.cells <- Cells(pbmc)Idents(object = pbmc, cells = pbmc.cells[1:10]) <- "CD4 T cells"#获取细胞标识类Idents(object = pbmc)levels(x = pbmc)#将细胞标识类存储在metadata中pbmc[["old.ident"]] <- Idents(object = pbmc)pbmc <- StashIdent(object = pbmc, save.name = "old.ident")#重命名标识类pbmc <- RenameIdents(object = pbmc, `CD4 T cells` = "T Helper cells")
3. 细胞metadata
#查看存储在object@meta.data中的meta.data数据框pbmc[[]]#从metadata中检索特定的值pbmc$nCount_RNApbmc[[c("percent.mito", "nFeature_RNA")]]#添加metadata，参见?AddMetaDatarandom_group_labels <- sample(x = c("g1", "g2"), size = ncol(x = pbmc), replace = TRUE)pbmc$groups <- random_group_labels
4. 表达式数据（在 Seurat v5 中作为layers存储）
#检索表达矩阵RNA计数矩阵中的数据pbmc[["RNA"]]$counts#具有相同结果的备用函数LayerData(pbmc, assay = "RNA", layer = "counts")#仍然支持来自Seurat v4的GetAssayDataGetAssayData(object = pbmc, assay = "RNA", slot = "counts")#ADT计数矩阵(多模态对象)cbmc[["ADT"]]$counts
#设置表达式数据假设为new.data 是一种新的表达式矩阵pbmc[["RNA"]]$counts <- new.data#具有相同结果的备用setter函数LayerData(pbmc, assay = "RNA", layer = "counts") <- new.data#仍然支持来自Seurat v4的SetAssayDatapbmc <- SetAssayData(object = pbmc, slot = "counts", new.data = new.data
5. 降维
#获取存储在pbmc上的细胞嵌入和基因加载[['pca']]@cell.embeddingsEmbeddings(pbmc, reduction = "pca")#存储在pbmc[['pca]]@feature.loadings中Loadings(pbmc, reduction = "pca")#创建自定义尺寸缩减加载矩阵是可选的new_reduction <- CreateDimReducObject(embeddings = new.embeddings, loadings = new.loadings, key = "custom_pca")pbmc[["custom_pca"]] <- new_reduction
6. FetchData （获取数据）
#FetchData可以访问表达式矩阵、细胞嵌入或metadata中的任何内容#使用FetchData访问单个/小组变量FetchData(object = pbmc, vars = c("PC_1", "nFeature_RNA", "MS4A1"), layer = "counts")
三、子集和合并
1. Seurat对象的子集
#基于标识类的子集Seurat对象，参见?SubsetDatasubset(x = pbmc, idents = "B")subset(x = pbmc, idents = c("Naive CD4 T", "CD8 T"), invert = TRUE)#基因/特征表达水平上的子集subset(x = pbmc, subset = MS4A1 > 2.5)#标准的组合上的子集subset(x = pbmc, subset = MS4A1 > 2.5 & PC_1 > 5)subset(x = pbmc, subset = MS4A1 > 2.5, idents = "B")#对象metadata中的子集subset(x = pbmc, subset = groups == "g1")#下采样每个标识类的细胞数量subset(x = pbmc, downsample = 100)
2. 分层
#在Seurat v5中，用户现在可以将对象直接拆分为不同的层，将表达式数据保留在一个对象中，但是#将多个样本拆分成层后可以直接进入整合工作流程ifnb[["RNA"]] <- split(ifnb[["RNA"]], f = ifnb$stim)Layers(ifnb)#如果需要，例如在整合之后，这些层可以再次连接在一起ifnb <- JoinLayers(ifnb)
3. 拆分对象
#可以根据metadata将对象拆分为多个对象的列表#创建两个对象的列表ifnb_list <- SplitObject(ifnb, split.by = "stim")ifnb_list$CTRLifnb_list$STIM
4. 合并对象（没有integration）
integration。如果不继续integration，请在合并后重新加入layers。#合并两个Seurat对象merged_obj <- merge(x = ifnb_list$CTRL, y = ifnb_list$STIM)merged_obj[["RNA"]] <- JoinLayers(merged_obj)#合并两个以上的Seurat对象merge(x = pbmc1, y = list(pbmc2, pbmc3))
5. 合并对象（有integration）
https://satijalab.org/seurat/articles/integration_introduction
merged_obj <- merge(x = ifnb_list$CTRL, y = ifnb_list$STIM)merged_obj <- NormalizeData(merged_obj)merged_obj <- FindVariableFeatures(merged_obj)merged_obj <- ScaleData(merged_obj)merged_obj <- RunPCA(merged_obj)merged_obj <- IntegrateLayers(object = obj, method = RPCAIntegration, orig.reduction = "pca", new.reduction = "integrated.rpca",verbose = FALSE)#现在整合已经完成，重新加入layersmerged_obj[["RNA"]] <- JoinLayers(merged_obj)
为了从高通量单细胞mRNA数据中增加基因数量和基因表达相关性，我们从同一细胞群中的多个细胞中收集数据，制作假细胞（Pseudocell）用于网络解释。看来这个Pseudocell概念是为了弥补稀疏矩阵在计算相关性上的缺陷，毕竟零值太多，影响相关性的计算。
基于多个类别将细胞分组在一起
有关如何将列添加到metadata的信息，请参阅https://satijalab.org/seurat/articles/de_vignette。donor_id
#pseudobulk细胞仅按细胞类型区分bulk <- AggregateExpression(ifnb, group.by = "seurat_annotations", return.seurat = TRUE)Cells(bulk)#pseudobulk细胞受刺激条件和细胞类型的影响bulk <- AggregateExpression(ifnb, group.by = c("stim", "seurat_annotations"), return.seurat = TRUE)Cells(bulk)#pseudobulk细胞与刺激条件、细胞类型和供体有关bulk <- AggregateExpression(ifnb, group.by = c("stim", "seurat_annotations", "donor_id"), return.seurat = TRUE)Cells(bulk)
Seurat 有一个庞大的、基于 ggplot2 的绘图库。默认情况下，所有绘图函数都将返回一个 ggplot2 绘图，从而允许使用 ggplot2 轻松自定义。
#降维图DimPlot(object = pbmc, reduction = "pca")#维度缩减图 细胞由定量特征着色 如果可用，默认为UMAPFeaturePlot(object = pbmc, features = "MS4A1")#单个细胞的散点图FeatureScatter(object = pbmc, feature1 = "MS4A1", feature2 = "PC_1")FeatureScatter(object = pbmc, feature1 = "MS4A1", feature2 = "CD3D")#散点图跨越各个特征 取代CellPlotCellScatter(object = pbmc, cell1 = "AGTCTACTAGGGTG", cell2 = "CACAGATGGTTTCT")VariableFeaturePlot(object = pbmc)#小提琴和山脊图VlnPlot(object = pbmc, features = c("LYZ", "CCL5", "IL32"))RidgePlot(object = pbmc, feature = c("LYZ", "CCL5", "IL32"))
#热图(可视化scale.data)DimHeatmap(object = pbmc, reduction = "pca", cells = 200)#标准工作流程pbmc <- ScaleData(pbmc, features = heatmap_markers)DoHeatmap(object = pbmc, features = heatmap_markers)#sctransform工作流程pbmc <- GetResidual(pbmc, features = heatmap_markers)DoHeatmap(object = pbmc, features = heatmap_markers)#热图 每组最多100个细胞DoHeatmap(pbmc, heatmap_markers, cells = subset(pbmc, downsample = 100)
#尝试新事物!注意，绘图函数现在返回ggplot2对象#因此可以向其添加主题、标题和选项VlnPlot(object = pbmc, features = "MS4A1", split.by = "groups")DotPlot(object = pbmc, features = c("LYZ", "CCL5", "IL32"), split.by = "groups")FeaturePlot(object = pbmc, features = c("MS4A1", "CD79A"), blend = TRUE)DimPlot(object = pbmc) + DarkTheme()DimPlot(object = pbmc) + labs(title = "2,700 PBMCs clustered using Seurat and viewed\non a two-dimensional UMAP")
Seurat 提供了许多预构建的主题，可以添加到 ggplot2 绘图中以便快速自定义。
| 主题 | 功能 |
|---|---|
DarkTheme | 设置带有白色文本的黑色背景 |
FontSize | 设置绘图中各种元素的字体大小 |
NoAxes | 删除轴和轴文本 |
NoLegend | 删除所有图例元素 |
RestoreLegend | 删除后恢复图例 |
RotatedAxis | 旋转 x 轴标签 |
#绘图辅助函数用于基于ggplot2的散点图#如DimPlot, FeaturePlot, CellScatter和FeatureScatterplot <- DimPlot(object = pbmc) + NoLegend()#HoverLocator取代了以前的`do.hover`#它还可以通过' information '参数显示额外的数据 旨在与FetchData一起顺利工作HoverLocator(plot = plot, information = FetchData(object = pbmc, vars = c("ident", "PC_1", "nFeature_RNA")))#FeatureLocator取代了之前的`do.identify`select.cells <- FeatureLocator(plot = plot)#在ggplot对象上标记点LabelPoints(plot = plot, points = TopCells(object = pbmc[["pca"]]), repel = TRUE)
使用Seurat，您可以轻松地在单细胞水平的不同检测方法之间切换（例如来自CITE-seq的ADT计数，或整合/批量校正的数据）。大多数函数现在都采用检测参数，但您可以设置默认检测以避免重复语句。
cbmc <- CreateSeuratObject(counts = cbmc.rna)#添加ADT数据cbmc[["ADT"]] <- CreateAssayObject(counts = cbmc.adt)#通过指定要使用的assay来运行分析NormalizeData(object = cbmc, assay = "RNA")NormalizeData(object = cbmc, assay = "ADT", method = "CLR")#检索并设置默认assayDefaultAssay(object = cbmc)DefaultAssay(object = cbmc) <- "ADT"DefaultAssay(object = cbmc)#使用keys从两个assays中提取特征表达式FetchData(object = cbmc, vars = c("rna_CD3E", "adt_CD3"))#使用keys绘制来自多个assays的数据FeatureScatter(object = cbmc, feature1 = "rna_CD3E", feature2 = "adt_CD3")
对 Seurat v5 中数据存储的一些技术更改特别感兴趣的用户可以浏览以下资源：