sceList.integrated <- IntegrateData(anchorset = sceList.anchors, dims = 1:30)
from：https://satijalab.org/seurat/articles/integration_large_datasets.html
對(duì)于非常大的數(shù)據(jù)集，可以采用兩個(gè)選項(xiàng)來(lái)提高效率和運(yùn)行時(shí)間
1.Reciprocal PCA (RPCA)
2.Reference-based integration
按照RPCA流程還是報(bào)錯(cuò)
features <- SelectIntegrationFeatures(object.list = sceList)
sceList <- lapply(X = sceList, FUN = function(x) {
? x <- ScaleData(x, features = features, verbose = FALSE)
? x <- RunPCA(x, features = features, verbose = FALSE)
})
#由于此數(shù)據(jù)集包含PD RD SD，因此我們將選擇2PR 2PD 2SD以在基于參考的工作流程中使用。
anchors <- FindIntegrationAnchors(object.list = sceList, reference = c(1,2,4,7,23,24), reduction = "rpca",
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? dims = 1:50)
sceList.integrated <- IntegrateData(anchorset = anchors, dims = 1:50)
按照RPCA流程，將參考數(shù)據(jù)改為1PD 1PD 1SD，運(yùn)行成功
anchors <- FindIntegrationAnchors(object.list = sceList, reference = c(1,4,23), reduction = "rpca",
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? dims = 1:50)
sceList.integrated <- IntegrateData(anchorset = anchors, dims = 1:50)
