{"id":432,"date":"2023-08-15T22:19:38","date_gmt":"2023-08-15T14:19:38","guid":{"rendered":"https:\/\/blog.ineuro.net\/?p=432"},"modified":"2023-09-08T13:26:39","modified_gmt":"2023-09-08T05:26:39","slug":"%e5%8d%95%e7%bb%86%e8%83%9e%e6%b5%8b%e5%ba%8f%ef%bc%88%e4%ba%8c%ef%bc%89%ef%bc%9a%e5%a4%9a%e7%bb%84%e5%8d%95%e7%bb%86%e8%83%9e%e6%95%b0%e6%8d%ae%e7%9a%84%e5%90%88%e5%b9%b6","status":"publish","type":"post","link":"https:\/\/blog.ineuro.net\/index.php\/2023\/08\/15\/432\/","title":{"rendered":"\u5355\u7ec6\u80de\u6d4b\u5e8f\uff08\u4e8c\uff09\uff1a\u591a\u7ec4\u5355\u7ec6\u80de\u6570\u636e\u7684\u5408\u5e76"},"content":{"rendered":"

\u5355\u7ec6\u80de\u7684\u6279\u6b21\u6548\u5e94<\/h1>\n

\u6279\u6b21\uff1a\u7ec6\u80de\u4ee5\u4e0d\u540c\u7684\u5206\u7ec4\u5904\u7406\u65f6\uff0c\u53ef\u80fd\u53d1\u751f\u6279\u6b21\u6548\u5e94\u3002\u4e0d\u540c\u5355\u7ec6\u80de\u6837\u672c\u7684\u6570\u636e\u5408\u5e76\u901a\u5e38\u6709\u591a\u79cd\u65b9\u5f0f\uff0c\u5305\u542b\u951a\u5b9a\u70b9\u6cd5\u3001Harmony\u3001RPCA\u548cSCTransform\u7b49\u3002\u53ef\u4ee5\u6839\u636e\u5b9e\u9645\u60c5\u51b5\u9009\u62e9\u9700\u8981\u7684\u5408\u5e76\u65b9\u5f0f\u3002\u4e0b\u9762\u5206\u522b\u4ecb\u7ecd\u51e0\u79cd\u6d88\u9664\u6279\u6b21\u6548\u5e94\u7684\u5408\u5e76\u65b9\u5f0f\u3002<\/p>\n

\u4e00\u3001\u951a\u5b9a\u70b9\u6cd5<\/h1>\n

1. \u603b\u4f53\u4ee3\u7801<\/h2>\n
###\u52a0\u8f7d\u6240\u9700\u8981\u7684\u5305\nlibrary(Seurat)\nlibrary(tidyverse)\nlibrary(dplyr)\nlibrary(patchwork)\n\nx =list.files()\n\ndir = c('BC2\/', "BC21\/")\nnames(dir) = c('BC2',  'BC21')      \ndir\ncounts <- Read10X(data.dir =dir)\nscRNA_m = CreateSeuratObject(counts,min.cells = 3, min.features = 200)\ntable(scRNA_m@meta.data$orig.ident)\ndir[1]\nscRNA_list <- SplitObject(scRNA_m, split.by = "orig.ident")\n?SplitObject\n\nscRNAlist <- list()\nfor(i in 1:length(dir)){\n  counts <- Read10X(data.dir = dir[i])\n  scRNAlist[[i]] <- CreateSeuratObject(counts, min.cells = 3, min.features =300)\n}\n\ncounts <- Read10X(data.dir = "BC2\/")\nscRNAlist[[1]] <- CreateSeuratObject(counts, min.cells = 3, min.features =300)\n\ncounts <- Read10X(data.dir = "BC3\/")\nscRNAlist[[2]] <- CreateSeuratObject(counts, min.cells = 3, min.features =300)\n\ncounts <- Read10X(data.dir = "BC5\/")\nscRNAlist[[3]] <- CreateSeuratObject(counts, min.cells = 3, min.features =300)\n\n##\u5f52\u4e00\u5316\u6570\u636e\nfor (i in 1:length(scRNA_list)) {\n  scRNA_list[[i]] <- NormalizeData(scRNA_list[[i]])\n  scRNA_list[[i]] <- FindVariableFeatures(scRNA_list[[i]], selection.method = "vst",nfeatures = 3000)\n}\n\nlibrary(future)\nplan("multicore", workers =128)\noptions(future.globals.maxSize = 2000 * 1024^2)\n?SelectIntegrationFeatures\n\nfeatures <- SelectIntegrationFeatures(object.list = scRNA_list)\n\n?FindIntegrationAnchors\n\nscRNA_anchors <- FindIntegrationAnchors(object.list = scRNA_list,anchor.features = features)\n\nscRNA_inte <- IntegrateData(anchorset = scRNA_anchors)\n\nsave(scRNA1,file = "IntegrateData.Rdata")\nscRNA_inte@\n\nx=scRNA1@meta.data\nz1=x$orig.ident\nz2=x[,1]\nz3=x[1,]\nclass(x)\ncharacter()\ndata.frame()\nmatrix()\nlist()\n\n?DefaultAssay\n\nDefaultAssay(scRNA_inte) <- "integrated"\n\nscRNA_inte_scale <- ScaleData(scRNA_inte)\n\nscRNA_inte_PCA <- RunPCA(scRNA_inte_scale, npcs = 30, verbose = T)\n\n# t-SNE and Clustering\n\nscRNA_inte_FN <- FindNeighbors(scRNA_inte_PCA, reduction = "pca", dims = 1:20)\nscRNA_inte_FC <- FindClusters(scRNA_inte_FN, resolution = 0.8)\nscRNA_UMAP <- RunUMAP(scRNA_inte_FC, reduction = "pca", dims = 1:20)\ncolnames(scRNA1@meta.data)\nscRNA_TSNE <- RunTSNE(scRNA_inte_FC, dims = 1:20)\ncolnames(scRNA1@meta.data)\nDimPlot(scRNA_UMAP, reduction = "umap", group.by = "orig.ident")\nDimPlot(scRNA_UMAP, reduction = "umap", label = TRUE)\n\nDimPlot(scRNA_TSNE, reduction = "tsne", group.by = "orig.ident")\nDimPlot(scRNA_TSNE, reduction = "tsne", label = TRUE)\n<\/code><\/pre>\n
2. SplitObject()<\/h2>\n
\u786e\u5b9a\u5404\u4e2a\u6570\u636e\u96c6\u4e4b\u95f4\u7684 "\u951a\u70b9"\u3002\u9996\u5148\uff0c\u5c06\u7ec4\u5408\u5bf9\u8c61\u62c6\u5206\u6210\u4e00\u4e2a\u5217\u8868\uff0c\u6bcf\u4e2a\u6570\u636e\u96c6\u90fd\u662f\u5176\u4e2d\u7684\u4e00\u4e2a\u5143\u7d20\uff08\u8fd9\u662f\u5fc5\u8981\u7684\uff0c\u56e0\u4e3a\u6570\u636e\u662f\u6346\u7ed1\u5728\u4e00\u8d77\u7684\uff0c\u4fbf\u4e8e\u5206\u53d1\uff09\u3002<\/p>\n
\u5728\u5bfb\u627e\u951a\u70b9\u4e4b\u524d\uff0c\u4f1a\u8fdb\u884c\u6807\u51c6\u7684\u9884\u5904\u7406\uff08\u5bf9\u6570\u6b63\u6001\u5316\uff09\uff0c\u5e76\u4e3a\u6bcf\u4e2a\u951a\u70b9\u5355\u72ec\u8bc6\u522b\u53d8\u91cf\u7279\u5f81\u3002\u8bf7\u6ce8\u610f\uff0cSeurat \u57fa\u4e8e\u65b9\u5dee\u7a33\u5b9a\u53d8\u6362\uff08"vst"\uff09\u5b9e\u73b0\u4e86\u4e00\u79cd\u6539\u8fdb\u7684\u53d8\u91cf\u7279\u5f81\u9009\u62e9\u65b9\u6cd5<\/p>\n
3.SelectIntegrationFeatures()<\/h2>\n
\u9009\u62e9\u5728\u4e0d\u540c\u6570\u636e\u96c6\u4e2d\u53cd\u590d\u53d8\u5316\u7684\u7279\u5f81\u8fdb\u884c\u6574\u5408.<\/p>\n
\u5728\u6574\u5408\u591a\u4e2a\u6570\u636e\u96c6\u65f6\u9009\u62e9\u8981\u4f7f\u7528\u7684\u7279\u5f81\u3002\u6b64\u51fd\u6570\u6839\u636e\u88ab\u8ba4\u4e3a\u662f\u53d8\u91cf\u7684\u6570\u636e\u96c6\u6570\u91cf\u5bf9\u7279\u5f81\u8fdb\u884c\u6392\u5e8f\uff0c\u5e76\u6839\u636e\u5404\u6570\u636e\u96c6\u53d8\u91cf\u7279\u5f81\u6392\u5e8f\u7684\u4e2d\u4f4d\u6570\u6253\u7834\u5e73\u5c40\u3002\u5b83\u4f1a\u6839\u636e\u8fd9\u4e00\u6392\u540d\u8fd4\u56de\u5f97\u5206\u6700\u9ad8\u7684\u7279\u5f81\u3002<\/p>\n
4. FindIntegrationAnchors()<\/h2>\n
\u4f7f\u7528 FindIntegrationAnchors() \u51fd\u6570\uff08\u8be5\u51fd\u6570\u5c06 Seurat \u5bf9\u8c61\u5217\u8868\u4f5c\u4e3a\u8f93\u5165\uff09\u8bc6\u522b\u951a\u70b9\uff0c\u5e76\u4f7f\u7528 IntegrateData() \u5c06\u4e24\u4e2a\u6570\u636e\u96c6\u6574\u5408\u5728\u4e00\u8d77\u3002<\/p>\n
5. DefaultAssay()<\/h2>\n
\u6307\u5b9a\u6211\u4eec\u5c06\u5bf9\u4fee\u6b63\u540e\u7684\u6570\u636e\u8fdb\u884c\u4e0b\u6e38\u5206\u6790\u3002\u539f\u59cb\u672a\u7ecf\u4fee\u6539\u7684\u6570\u636e\u4ecd\u4fdd\u7559\u5728 "RNA "\u68c0\u6d4b\u4e2d\u3002<\/p>\n
\u4e8c\u3001Harmony<\/h1>\n
1.Harmony\u603b\u4f53\u4f18\u52bf<\/h2>\n
\u5b98\u65b9\u7f51\u7ad9\uff1ahttps:\/\/github.com\/immunogenomics\/harmony<\/a><\/p>\n
\uff081\uff09\u6574\u5408\u6570\u636e\u7684\u540c\u65f6\u5bf9\u7a00\u6709\u7ec6\u80de\u5b58\u5728\u4e00\u5b9a\u7684\u654f\u611f\u6027\uff1b<\/p>\n
\uff082\uff09\u7701\u5185\u5b58\uff1b<\/p>\n
\uff083\uff09\u9002\u5408\u4e8e\u590d\u6742\u7684\u5355\u7ec6\u80de\u5206\u6790\u8bbe\u8ba1\uff0c\u53ef\u4ee5\u6bd4\u8f83\u4e0d\u540c\u5e73\u53f0\u3001\u7ec4\u7ec7\u548c\u4f9b\u4f53\u7684\u7ec6\u80de\u3002<\/p>\n
2. \u603b\u4f53\u4ee3\u7801<\/h2>\n
DefaultAssay(scRNA_inte) <- "RNA"\nscRNA_harmony <- ScaleData(scRNA_inte)\n##\u5b89\u88c5Harmony\u5305\n\ninstall.packages("devtools")\nlibrary(devtools)\ninstall_github("immunogenomics\/harmony")\nlibrary(harmony)\nBiocManager::install("SingleCellExperiment")\n\nscRNA_1 <- Read10X("BC2\/")\nscRNA_2 <- Read10X("BC21\/")\nscRNA_1 <- CreateSeuratObject(scRNA_1 ,project="sample_1",min.cells = 3, min.features = 200)\nscRNA_2 <- CreateSeuratObject(scRNA_2 ,project="sample_2",min.cells = 3, min.features = 200)\n?CreateSeuratObject\n?merge\nscRNA_harmony <- merge(scRNA_1, y=c(scRNA_2 ))\nscRNA_harmony <- NormalizeData(scRNA_harmony) %>% FindVariableFeatures() %>% ScaleData() %>% RunPCA(verbose=FALSE)\n?system.time\n?RunHarmony\nsystem.time({scRNA_harmony <- RunHarmony(scRNA_harmony, group.by.vars = "orig.ident")})\n\n#\u6307\u5b9aharmony\nscRNA_harmony <- FindNeighbors(scRNA_harmony, reduction = "harmony", dims = 1:15) %>% FindClusters(resolution = 0.5)\n\nscRNA_harmony <- RunUMAP(scRNA_harmony, reduction = "harmony", dims = 1:16)\n\nplot1 =DimPlot(scRNA_harmony, reduction = "umap",label = T) \nplot2 = DimPlot(scRNA_harmony, reduction = "umap", group.by='orig.ident') \n#combinate\nplot_c <- plot1+plot2\nplot_c<\/code><\/pre>\n
3. merge()<\/h2>\n
\u901a\u8fc7\u5171\u540c\u5217\u540d\u6216\u884c\u540d\u5408\u5e76\u4e24\u4e2a\u6570\u636e\u5e27\uff0c\u6216\u6267\u884c\u5176\u4ed6\u7248\u672c\u7684\u6570\u636e\u5e93 join<\/em> \u64cd\u4f5c]\u3002<\/p>\n
4. \u7ba1\u9053\u7b26<\/h2>\n
\u7ba1\u9053\u64cd\u4f5c\u7b26\uff08Pipe Operator\uff09\u662f\u4e00\u4e2a\u7279\u5b9a\u7684\u7b26\u53f7\uff0c\u5b83\u53ef\u4ee5\u5c06\u524d\u4e00\u884c\u4ee3\u7801\u7684\u8f93\u51fa\u4f20\u9012\u7ed9\u540e\u4e00\u884c\u4ee3\u7801\u4f5c\u4e3a\u8f93\u5165\uff0c\u4ece\u800c\u5c06\u539f\u672c\u76f8\u4e92\u72ec\u7acb\u7684\u4e24\u884c\u4ee3\u7801\u8fde\u63a5\u5728\u4e00\u8d77\u3002\u800c\u901a\u8fc7\u4e0d\u65ad\u5730\u4f7f\u7528\u7ba1\u9053\u64cd\u4f5c\u7b26\uff0c\u6700\u7ec8\u53ef\u4ee5\u5c06\u591a\u884c\u4ee3\u7801\u5199\u6210\u201c\u6d41\u201d\u7684\u5f62\u5f0f\u3002\u4f7f\u7528\u7ba1\u9053\u64cd\u4f5c\u7b26\u65e2\u53ef\u4ee5\u7b80\u5316\u4ee3\u7801\uff0c\u53c8\u53ef\u4ee5\u4f7f\u4ee3\u7801\u95f4\u7684\u903b\u8f91\u5173\u7cfb\u66f4\u52a0\u6e05\u6670\uff0c\u8fd8\u53ef\u4ee5\u7701\u53bb\u4e2d\u95f4\u53d8\u91cf\u7684\u8f93\u51fa\u3002<\/p>\n
%>%<\/h3>\n
\u5982\u679c\u4e00\u884c\u4ee3\u7801\u9700\u8981\u8f93\u5165\u7684\u53c2\u6570\u503c\u521a\u597d\u662f\u5b83\u524d\u4e00\u884c\u7684\u8f93\u51fa\u7ed3\u679c\uff0c\u53ef\u4ee5\u4f7f\u7528%>%<\/code>\u7701\u7565\u4e2d\u95f4\u7684\u8f93\u5165\u8fc7\u7a0b\u3002\u5982\u679c\u53c2\u6570\u662f\u4f4d\u4e8e\u7b2c\u4e00\u7684\u4f4d\u7f6e\u53ef\u4ee5\u76f4\u63a5\u7701\u7565\uff08\u5927\u591a\u6570\u90fd\u662f\u8fd9\u79cd\u60c5\u51b5\uff09\uff0c\u5176\u4ed6\u4f4d\u7f6e\u7684\u53c2\u6570\u7167\u5e38\u4e66\u5199\u3002\u53c2\u6570\u4e0d\u662f\u4f4d\u4e8e\u7b2c\u4e00\u7684\u4f4d\u7f6e\uff0c\u5219\u9700\u8981\u989d\u5916\u4f7f\u7528\u5360\u4f4d\u7b26\u3002<\/p>\n
\u5f53\u6709\u591a\u4e2a\u53c2\u6570\u7684\u8f93\u5165\u503c\u4f9d\u8d56\u4e8e\u524d\u9762\u4ee3\u7801\u7684\u8f93\u51fa\u7ed3\u679c\u65f6\uff0c\u9700\u8981\u7ed3\u5408\u5927\u62ec\u53f7{}\u8fdb\u884c\u4f7f\u7528<\/strong><\/p>\n
# \u4e0d\u4f7f\u7528\u7ba1\u9053\u64cd\u4f5c\u7b26\ndata <- mtcars[,1]\nn <- length(data)\nmin <- min(data)\nmax <- max(data)\nrdn <- runif(n = n, min = min, max = max)\n\n# \u4f7f\u7528\u7ba1\u9053\u64cd\u4f5c\u7b26\nmtcars %>%\n  pluck(1) %>%\n  {\n    n = min(.)\n    min = min(.)\n    max = max(.)\n    runif(n = n, min = min, max = max)\n  } -> rdn\n\n\u539f\u6587\u94fe\u63a5\uff1ahttps:\/\/blog.csdn.net\/weixin_54000907\/article\/details\/114481943<\/code><\/pre>\n
5. system.time()<\/h2>\n
system.time()<\/code> \u51fd\u6570\u662f\u6211\u4eec\u53ef\u4ee5\u7528\u6765\u4f30\u8ba1\u4ee3\u7801\u6267\u884c\u6240\u9700\u65f6\u95f4\u7684\u5de5\u5177\u4e4b\u4e00\u3002\u5b83\u7684\u8f93\u51fa\u7ed9\u51fa\u4e86\u4e09\u4e2a\u503c\uff1a\u7528\u6237\u3001\u7cfb\u7edf\u548c\u7ecf\u8fc7\u7684\u65f6\u95f4\uff08\u4ee5\u79d2\u4e3a\u5355\u4f4d\uff09\u3002\u7528\u6237\u65f6\u95f4\u662f\u5904\u7406\u7528\u6237\u5e94\u7528\u7a0b\u5e8f\u4ee3\u7801\uff08R \u4ee3\u7801\uff09\u6240\u82b1\u8d39\u7684 CPU \u65f6\u95f4\u3002\u5982\u679c\u7528\u6237\u5e94\u7528\u7a0b\u5e8f\u8bbf\u95ee\u7cfb\u7edf\u8d44\u6e90\uff0c\u5219\u8be5\u5904\u7406\u65f6\u95f4\u5c06\u62a5\u544a\u4e3a\u7cfb\u7edf\u65f6\u95f4\u3002<\/p>\n
6. RunHarmony()<\/h2>\n
\u4f7f\u7528 Seurat \u548c SingleCellAnalysis \u6d41\u7a0b\u8fd0\u884cHarmony\u7b97\u6cd5\u3002<\/p>\n
\u4e09\u3001RPCA<\/h1>\n
\u4e00\u4e2a\u7a0d\u4f5c\u4fee\u6539\u7684\u5de5\u4f5c\u6d41\u7a0b\uff0c\u7528\u4e8e\u6574\u5408 scRNA-seq \u6570\u636e\u96c6\u3002\u4e0d\u4f7f\u7528\u5178\u578b\u76f8\u5173\u5206\u6790\uff08CCA\uff09\u6765\u786e\u5b9a\u951a\u70b9\uff0c\u800c\u662f\u4f7f\u7528\u76f8\u4e92PCA\uff08RPCA\uff09\u3002\u5728\u4f7f\u7528 RPCA \u786e\u5b9a\u4efb\u610f\u4e24\u4e2a\u6570\u636e\u96c6\u4e4b\u95f4\u7684\u951a\u70b9\u65f6\uff0c\u6211\u4eec\u4f1a\u5c06\u6bcf\u4e2a\u6570\u636e\u96c6\u6295\u5f71\u5230\u53e6\u4e00\u4e2a PCA \u7a7a\u95f4\uff0c\u5e76\u6839\u636e\u76f8\u540c\u7684\u4e92\u90bb\u8981\u6c42\u5bf9\u951a\u70b9\u8fdb\u884c\u7ea6\u675f\u3002\u8fd9\u4e24\u79cd\u5de5\u4f5c\u6d41\u7a0b\u7684\u547d\u4ee4\u57fa\u672c\u76f8\u540c\uff0c\u4f46\u8fd9\u4e24\u79cd\u65b9\u6cd5\u53ef\u4ee5\u5728\u4e0d\u540c\u7684\u73af\u5883\u4e2d\u5e94\u7528\u3002<\/p>\n
1. \u603b\u4f53\u4ee3\u7801<\/h2>\n
# split the dataset into a list of two seurat objects (stim and CTRL)\nifnb.list <- SplitObject(ifnb, split.by = "stim")\n\n# normalize and identify variable features for each dataset independently\nifnb.list <- lapply(X = ifnb.list, FUN = function(x) {\n    x <- NormalizeData(x)\n    x <- FindVariableFeatures(x, selection.method = "vst", nfeatures = 2000)\n})\n\n# select features that are repeatedly variable across datasets for integration run PCA on each\n# dataset using these features\nfeatures <- SelectIntegrationFeatures(object.list = ifnb.list)\nifnb.list <- lapply(X = ifnb.list, FUN = function(x) {\n    x <- ScaleData(x, features = features, verbose = FALSE)\n    x <- RunPCA(x, features = features, verbose = FALSE)\n})\n##\u5148\u7f29\u653e\u6570\u636e\u3001\u964d\u7ef4\uff0c\u540e\u5bfb\u627e\u951a\u5b9a\u70b9\nimmune.anchors <- FindIntegrationAnchors(object.list = ifnb.list, anchor.features = features, reduction = "rpca")\n#\u6574\u5408\u6570\u636e\uff0c\u540e\u9762\u7684\u4ee3\u7801\u5747\u4e0e\u951a\u5b9a\u70b9\u6cd5\u76f8\u540c\nimmune.combined <- IntegrateData(anchorset = immune.anchors)\n# specify that we will perform downstream analysis on the corrected data note that the\n# original unmodified data still resides in the 'RNA' assay\nDefaultAssay(immune.combined) <- "integrated"\n\n# Run the standard workflow for visualization and clustering\nimmune.combined <- ScaleData(immune.combined, verbose = FALSE)\nimmune.combined <- RunPCA(immune.combined, npcs = 30, verbose = FALSE)\nimmune.combined <- RunUMAP(immune.combined, reduction = "pca", dims = 1:30)\nimmune.combined <- FindNeighbors(immune.combined, reduction = "pca", dims = 1:30)\nimmune.combined <- FindClusters(immune.combined, resolution = 0.5)\n# Visualization\np1 <- DimPlot(immune.combined, reduction = "umap", group.by = "stim")\np2 <- DimPlot(immune.combined, reduction = "umap", group.by = "seurat_annotations", label = TRUE,\n    repel = TRUE)\np1 + p2\n<\/code><\/pre>\n
2. lapply()<\/h2>\n
lapply<\/code>\u662f\u4e3b\u529b\u7684\u51fd\u6570\u3002\u4ed6\u7684\u4e3b\u8981\u7528\u9014\u662f\uff0c\u5bf9\u5217\u8868<\/strong>\uff08list\uff09\u5bf9\u8c61\u800c\u8a00\uff0c\u4f60\u60f3\u5728\u5176\u5185\u90e8\u505a\u4e00\u4e2a\u5faa\u73af\uff0c\u5e76\u5bf9\u5217\u8868\u4e2d\u7684\u6bcf\u4e00\u4e2a\u5143\u7d20\u8fd0\u7528\u51fd\u6570\u3002<\/p>\n
sapply<\/code>\u662flapply<\/code>\u7684\u4e00\u4e2a\u53d8\u4f53\uff0c\u7b80\u5316\u4e86lapply<\/code>\u7684\u7ed3\u679c<\/p>\n
apply<\/code>\u662f\u4e00\u4e2a\u5bf9\u6570\u7ec4<\/strong>\u8fdb\u884c\u884c\u6216\u5217\u8fd0\u7b97\u7684\u51fd\u6570\u3002\u5982\u679c\u4f60\u60f3\u5bf9\u77e9\u9635\u6216\u5176\u4ed6\u9ad8\u7ef4\u6570\u7ec4\u6c42\u548c\uff0c\u8fd9\u4e2a\u51fd\u6570\u4f1a\u975e\u5e38\u597d\u7528\u3002<\/p>\n
tapply<\/code>\u662f`table apply()\u7684\u7f29\u5199\uff0c\u5c06\u51fd\u6570\u5e94\u7528\u4e8e\u5411\u91cf\u7684\u5b50\u96c6\u3002<\/p>\n
mapply<\/code>\u662f`apply\u7684\u591a\u53d8\u91cf\u7248\u672c\u3002<\/p>\n
3. \u66f4\u6539\u6574\u5408\u5f3a\u5ea6<\/h2>\n
\u57fa\u4e8e rpca \u7684\u6574\u5408\u66f4\u4e3a\u4fdd\u5b88\uff0c\u5728\u8fd9\u79cd\u60c5\u51b5\u4e0b\uff0c\u7ec6\u80de\u5b50\u96c6\uff08\u5373\u5e7c\u7a1a\u548c\u8bb0\u5fc6 T \u7ec6\u80de\uff09\u5728\u4e0d\u540c\u5b9e\u9a8c\u4e2d\u7684\u914d\u51c6\u5e76\u4e0d\u5b8c\u7f8e\u3002\u60a8\u53ef\u4ee5\u901a\u8fc7\u589e\u52a0\u9ed8\u8ba4\u8bbe\u7f6e\u4e3a 5 \u7684 k.anchor \u53c2\u6570\u6765\u63d0\u9ad8\u914d\u51c6\u5f3a\u5ea6\u3002\u5c06\u8be5\u53c2\u6570\u63d0\u9ad8\u5230 20 \u5c06\u6709\u52a9\u4e8e\u8fd9\u4e9b\u7ec6\u80de\u7fa4\u7684\u914d\u51c6\u3002<\/p>\n
immune.anchors <- FindIntegrationAnchors(object.list = ifnb.list, anchor.features = features, reduction = "rpca",\n    k.anchor = 20)\nimmune.combined <- IntegrateData(anchorset = immune.anchors)\n\nimmune.combined <- ScaleData(immune.combined, verbose = FALSE)\nimmune.combined <- RunPCA(immune.combined, npcs = 30, verbose = FALSE)\nimmune.combined <- RunUMAP(immune.combined, reduction = "pca", dims = 1:30)\nimmune.combined <- FindNeighbors(immune.combined, reduction = "pca", dims = 1:30)\nimmune.combined <- FindClusters(immune.combined, resolution = 0.5)<\/code><\/pre>\n
\u56db\u3001SCTtransform<\/h1>\n
SCTransform\u53ef\u4ee5\u7528\u4e8e\u6570\u636e\u6807\u51c6\u5316\uff0c\u4ee3\u66ff\u4e09\u4e2a\u51fd\u6570\uff08NormalizeData, ScaleDate, FindVariableFeatures\uff09\u7684\u8fd0\u884c\u3002\u5176\u5bf9\u6d4b\u5e8f\u6df1\u5ea6\u7684\u8f83\u6b63\u6548\u679c\u597d\u4e8elog\u6807\u51c6\u5316\u3002\uff0810\u4e07\u4ee5\u5185\u7684\u7ec6\u80de\u5747\u5efa\u8bae\u4f7f\u7528SCT\uff09\u3002\u4e0d\u80fd\u7528\u4e8e\u6279\u6b21\u8f83\u6b63\u3002<\/p>\n
SCTranform\u53ef\u4ee5\u9009\u62e9\u66f4\u591a\u7684\u4e3b\u6210\u5206\uff08PC\uff09\u3002\u4f7f\u7528SCTranform\u5728\u7b97\u6cd5\u4e0a\u66f4\u52a0\u201c\u81ea\u4fe1\u201d\uff0c\u53ef\u4ee5\u63d0\u53d6\u66f4\u591a\u751f\u7269\u5dee\u5f02\u4fe1\u606f\u3002<\/p>\n
\u9ad8\u53d8\u57fa\u56e0\u9009\u62e9\uff1a\u9ed8\u8ba4\u9009\u62e93000\u4e2a\u57fa\u56e0\uff0c\u53ef\u4ee5\u6db5\u76d6\u4e4b\u524d\u672a\u9009\u62e9\u7684\u57fa\u56e0\u3002<\/p>\n
1. \u5355\u4e2a\u6837\u672c\u7684\u5f52\u4e00\u5316<\/h2>\n
scRNA.counts=Read10X("BC21\/")\n\n###\u521b\u5efaSeurat\u5bf9\u8c61\nscRNA = CreateSeuratObject(scRNA.counts ,min.cells = 3,project="os", min.features = 300)\nscRNA <- PercentageFeatureSet(scRNA, pattern = "^MT-", col.name = "percent.mt")\nscRNA <- SCTransform(scRNA, vars.to.regress = "percent.mt", verbose = FALSE)\n\n######\u522b\u770bSCTransform\u53ea\u6709\u4e00\u4e2a\u5355\u72ec\u7684\u51fd\u6570\uff0c\u5176\u5b9e\u5b83\u505a\u4e86\uff1aNormalizeData \u3001ScaleData\u3001FindVariableFeatures \u7684\u4e8b\u60c5\n########\u5e76\u4e14\u4e5f\u652f\u6301ScaleData\u7684vars.to.regress\n\n##\u8fd0\u884c\u7684\u7ed3\u679c\u5b58\u50a8\u5728\uff1a\nscRNA@assays$SCT) \n#\u6216\u8005\nscRNA[["SCT"]]\n\n##\u964d\u7ef4\uff0c\u7136\u540e\u805a\u7c7b\n\nscRNA <- RunPCA(scRNA, verbose = FALSE)\nscRNA <- RunUMAP(scRNA, dims = 1:30, verbose = FALSE)\n\nscRNA <- FindNeighbors(scRNA, dims = 1:30, verbose = FALSE)\nscRNA <- FindClusters(scRNA, verbose = FALSE)\nDimPlot(scRNA, label = TRUE) + NoLegend()<\/code><\/pre>\n
2. \u591a\u4e2a\u6837\u672c\u5229\u7528SCT\u6574\u5408\uff08\u975eRPCA\u6cd5\uff09<\/h2>\n
\u5728 Hafemeister \u548c Satija 2019 \u5e74\u7684\u6587\u7ae0\u4e2d\uff0c\u4ecb\u7ecd\u4e86\u4e00\u79cd\u57fa\u4e8e\u6b63\u5219\u5316\u8d1f\u4e8c\u9879\u5f0f\u56de\u5f52\u7684 scRNA-seq \u5f52\u4e00\u5316\u6539\u8fdb\u65b9\u6cd5\u3002\u8be5\u65b9\u6cd5\u88ab\u547d\u540d\u4e3a "sctransform"\uff0c\u907f\u514d\u4e86\u6807\u51c6\u5f52\u4e00\u5316\u5de5\u4f5c\u6d41\u7a0b\u7684\u4e00\u4e9b\u7f3a\u9677\uff0c\u5305\u62ec\u6dfb\u52a0\u4f2a\u8ba1\u6570\u548c\u5bf9\u6570\u53d8\u6362\u3002<\/p>\n
\u5728\u6574\u5408\u4e4b\u524d\uff0c\u901a\u8fc7 SCTransform() \u800c\u4e0d\u662f NormalizeData() \u5bf9\u6570\u636e\u96c6\u8fdb\u884c\u5355\u72ec\u7684\u89c4\u8303\u5316\u5904\u7406
\n\u6b63\u5982\u6211\u4eec\u5728 SCTransform vignette \u4e2d\u8fdb\u4e00\u6b65\u8ba8\u8bba\u7684\uff0c\u6211\u4eec\u901a\u5e38\u4f7f\u7528 3,000 \u4e2a\u6216\u66f4\u591a\u7279\u5f81\u8fdb\u884c sctransform \u4e0b\u6e38\u5206\u6790\u3002
\n\u5728\u8bc6\u522b\u951a\u70b9\u524d\u8fd0\u884c PrepSCTIntegration() \u51fd\u6570
\n\u8fd0\u884c FindIntegrationAnchors() \u548c IntegrateData() \u65f6\uff0c\u5c06 normalization.method \u53c2\u6570\u8bbe\u7f6e\u4e3a SCT \u503c\u3002
\n\u8fd0\u884c\u57fa\u4e8e sctransform \u7684\u5de5\u4f5c\u6d41\uff08\u5305\u62ec\u96c6\u6210\uff09\u65f6\uff0c\u8bf7\u52ff\u8fd0\u884c ScaleData() \u51fd\u6570\u3002<\/p>\n
LoadData("ifnb")\nifnb.list <- SplitObject(ifnb, split.by = "stim")\nifnb.list <- lapply(X = ifnb.list, FUN = SCTransform)\nfeatures <- SelectIntegrationFeatures(object.list = ifnb.list, nfeatures = 3000)\nifnb.list <- PrepSCTIntegration(object.list = ifnb.list, anchor.features = features)\nimmune.anchors <- FindIntegrationAnchors(object.list = ifnb.list, normalization.method = "SCT",\n    anchor.features = features)\nimmune.combined.sct <- IntegrateData(anchorset = immune.anchors, normalization.method = "SCT")\nimmune.combined.sct <- RunPCA(immune.combined.sct, verbose = FALSE)\nimmune.combined.sct <- RunUMAP(immune.combined.sct, reduction = "pca", dims = 1:30)\np1 <- DimPlot(immune.combined.sct, reduction = "umap", group.by = "stim")\np2 <- DimPlot(immune.combined.sct, reduction = "umap", group.by = "seurat_annotations", label = TRUE,\n    repel = TRUE)\np1 + p2<\/code><\/pre>\n
3. \u591a\u4e2a\u6837\u672c\u5229\u7528SCT\u6574\u5408\uff08RPCA\u6cd5\uff09<\/h2>\n
SCTransform \u5bf9\u6570\u636e\u96c6\u8fdb\u884c\u5f52\u4e00\u5316\u5904\u7406\u3002\u6211\u4eec\u53ef\u4ee5\u9009\u62e9\u5c06\u65b9\u6cd5\u53c2\u6570\u8bbe\u7f6e\u4e3a glmGamPoi\uff0c\u4ee5\u4fbf\u5728 SCTransform() \u4e2d\u66f4\u5feb\u5730\u4f30\u8ba1\u56de\u5f52\u53c2\u6570\u3002<\/p>\n
LoadData("ifnb")\nifnb.list <- SplitObject(ifnb, split.by = "stim")\nifnb.list <- lapply(X = ifnb.list, FUN = SCTransform, method = "glmGamPoi")\nfeatures <- SelectIntegrationFeatures(object.list = ifnb.list, nfeatures = 3000)\nifnb.list <- PrepSCTIntegration(object.list = ifnb.list, anchor.features = features)\nifnb.list <- lapply(X = ifnb.list, FUN = RunPCA, features = features)\nimmune.anchors <- FindIntegrationAnchors(object.list = ifnb.list, normalization.method = "SCT",\n    anchor.features = features, dims = 1:30, reduction = "rpca", k.anchor = 20)\nimmune.combined.sct <- IntegrateData(anchorset = immune.anchors, normalization.method = "SCT", dims = 1:30)\nimmune.combined.sct <- RunPCA(immune.combined.sct, verbose = FALSE)\nimmune.combined.sct <- RunUMAP(immune.combined.sct, reduction = "pca", dims = 1:30)\n# Visualization\np1 <- DimPlot(immune.combined.sct, reduction = "umap", group.by = "stim")\np2 <- DimPlot(immune.combined.sct, reduction = "umap", group.by = "seurat_annotations", label = TRUE,\n    repel = TRUE)\np1 + p2<\/code><\/pre>\n","protected":false},"excerpt":{"rendered":"
\u5355\u7ec6\u80de\u7684\u6279\u6b21\u6548\u5e94 \u6279\u6b21\uff1a\u7ec6\u80de\u4ee5\u4e0d\u540c\u7684\u5206\u7ec4\u5904\u7406\u65f6\uff0c\u53ef\u80fd\u53d1\u751f\u6279\u6b21\u6548\u5e94\u3002\u4e0d\u540c\u5355\u7ec6\u80de\u6837\u672c\u7684\u6570\u636e\u5408\u5e76\u901a\u5e38\u6709\u591a\u79cd\u65b9\u5f0f\uff0c\u5305\u542b […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1,24],"tags":[],"topics":[],"_links":{"self":[{"href":"https:\/\/blog.ineuro.net\/index.php\/wp-json\/wp\/v2\/posts\/432"}],"collection":[{"href":"https:\/\/blog.ineuro.net\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blog.ineuro.net\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blog.ineuro.net\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/blog.ineuro.net\/index.php\/wp-json\/wp\/v2\/comments?post=432"}],"version-history":[{"count":5,"href":"https:\/\/blog.ineuro.net\/index.php\/wp-json\/wp\/v2\/posts\/432\/revisions"}],"predecessor-version":[{"id":437,"href":"https:\/\/blog.ineuro.net\/index.php\/wp-json\/wp\/v2\/posts\/432\/revisions\/437"}],"wp:attachment":[{"href":"https:\/\/blog.ineuro.net\/index.php\/wp-json\/wp\/v2\/media?parent=432"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blog.ineuro.net\/index.php\/wp-json\/wp\/v2\/categories?post=432"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blog.ineuro.net\/index.php\/wp-json\/wp\/v2\/tags?post=432"},{"taxonomy":"topics","embeddable":true,"href":"https:\/\/blog.ineuro.net\/index.php\/wp-json\/wp\/v2\/topics?post=432"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}