Azimuth 是Seurat那个组出的注释工具。 有网页版,有R包。
Azimuth annotation ? Azimuth (satijalab.github.io)
官网 (hubmapconsortium.org)
一、网页版 参考:单细胞分析之细胞注释-1:Azimuth - 简书 (jianshu.com)
将seurat对象的count矩阵保存为rds文件,直接输入Azimuth网站进行预测。然后下载Predicted cell types and scores (TSV),将下载得到的结果(azimuth_pred)重新上传到R中。查看Azimuth注释结果。
tmp <- scRNA1@assays$RNA@counts
saveRDS(tmp,file = 'scRNA1_count_0505_0.4.rds')
predictions = read.delim('azimuth_pred.tsv', row.names = 1)
scRNA1 = AddMetaData(scRNA1, metadata = predictions)
DimPlot(scRNA1, reduction = 'umap', group.by = "predicted.celltype.l2" , label=T)
试了下网页版Azimuth: Human PBMC (hubmapconsortium.org),得到是这样的?
最后的得分太低,应该是不行。试试R包吧。得分要在3-5是可信的。
后来又试了一次,得分是2点几,下载了Predicted cell types and scores (TSV) 试了下,结果是这样
二、R包
Azimuth annotation ? Azimuth (satijalab.github.io)
github
### Azimuth annotation https://satijalab.github.io/azimuth/articles/run_azimuth_tutorial.html
devtools::install_github("satijalab/seurat", "seurat5")
devtools::install_github("satijalab/seurat-data", "seurat5")
devtools::install_github("satijalab/azimuth", build = FALSE,force = TRUE)
library(Seurat)
library(Azimuth)
library(SeuratData)
library(patchwork)
# Install the PBMC systematic comparative analyis (pmbcsca) dataset
InstallData("pbmcsca")
# returns a Seurat object named pbmcsca
pbmcsca <- LoadData("pbmcsca")
# The RunAzimuth function can take a Seurat object as input
scRNA1Azi <- RunAzimuth(scRNA1, reference = "pbmcref")
p1 <- DimPlot(scRNA1Azi, group.by = "predicted.celltype.l2", label = TRUE, label.size = 3) + NoLegend()
p2 <- DimPlot(scRNA1Azi, group.by = 'orig.ident')
p1 + p2
结论:R包和网页版注释的结果一模一样。并且不是按照原来的分群去注释的。不知道是不是因为用harmony去批次,没有用Seurat去批次的原因。还是说这个包他就是给个大概的注释参考,不是完全给你注释好。
贴一下分群umap: