3.15 不同R包数据的相互转换

刘小泽写于2020.7.18

1 前言

这部分内容是来自Seurat：https://satijalab.org/seurat/v3.1/conversion_vignette.html

单细胞数据格式目前有这么几大派：

• Bioconductor主导的SingleCellExperiment数据格式：例如scran、scater、monocle（尽管它的对象不直接使用SingleCellExperiment，但灵感来源于SingleCellExperiment，并且操作也是类似的）

• Seurat：SeuratObject格式

• scanpy：AnnData格式

这么一来，很多分析流程就被固定在某个包中了，比如使用Seurat会一用到底，也不会去学习scater或其他R包了，但也许就错过了其他R包好用的一些功能（比如我感觉scater的uniquifyFeatureNames就很好用）

既然有需求，就有开发者添加功能 ，这里Davis McCarthy 和Alex Wolf就为Seurat添加了和其他数据类型转换的函数

2 Seurat与SingleCellExperiment的相互转换

library(scater)
# devtools::install_github(repo = "satijalab/seurat", ref = "loom")
library(loomR)
library(Seurat)
library(patchwork)

2.1 Seurat转SingleCellExperiment

# 使用Seurat内置数据
data("pbmc_small")
> pbmc_small
An object of class Seurat 
230 features across 80 samples within 1 assay 
Active assay: RNA (230 features)
 2 dimensional reductions calculated: pca, tsne

# 一个函数即可
pbmc.sce <- as.SingleCellExperiment(pbmc_small)
> pbmc.sce
class: SingleCellExperiment 
dim: 230 80 
metadata(0):
assays(2): counts logcounts
rownames(230): MS4A1 CD79B ... SPON2 S100B
rowData names(5): vst.mean vst.variance
  vst.variance.expected
  vst.variance.standardized vst.variable
colnames(80): ATGCCAGAACGACT CATGGCCTGTGCAT ...
  GGAACACTTCAGAC CTTGATTGATCTTC
colData names(8): orig.ident nCount_RNA ...
  RNA_snn_res.1 ident
reducedDimNames(2): PCA TSNE
spikeNames(0):
altExpNames(0):

# 接下来就是scater的操作了
p1 <- plotExpression(pbmc.sce, features = "MS4A1", x = "ident") + theme(axis.text.x = element_text(angle = 45, 
    hjust = 1))
p2 <- plotPCA(pbmc.sce, colour_by = "ident")
p1 + p2

2.2 SingleCellExperiment转Seurat

# 导入sce对象（https://scrnaseq-public-datasets.s3.amazonaws.com/scater-objects/manno_human.rds）
manno <- readRDS(file = "manno_human.rds")
> manno
class: SingleCellExperiment 
dim: 20560 4029 
metadata(0):
assays(2): counts logcounts
rownames(20560): 'MARC1' 'MARC2' ... ZZEF1 ZZZ3
rowData names(10): feature_symbol
  is_feature_control ... total_counts
  log10_total_counts
colnames(4029): 1772122_301_C02 1772122_180_E05
  ... 1772116-063_G02 1772099-259_H03
colData names(34): Species cell_type1 ...
  pct_counts_ERCC is_cell_control
reducedDimNames(0):
altExpNames(0):

manno <- runPCA(manno)
# 转为seurat对象
manno.seurat <- as.Seurat(manno, counts = "counts", data = "logcounts")

# 看下这个函数
# as.Seurat(
#     x,
#     counts = "counts",
#     data = "logcounts",
#     assay = "RNA",
#     project = "SingleCellExperiment",
#     ...
# )
# 既然有默认参数，因此直接按下面这么写就可以：
manno.seurat <- as.Seurat(manno)

> manno.seurat
An object of class Seurat 
20560 features across 4029 samples within 1 assay 
Active assay: RNA (20560 features)
 1 dimensional reduction calculated: PCA

Idents(manno.seurat) <- "cell_type1"
p1 <- DimPlot(manno.seurat, reduction = "PCA", group.by = "Source") + NoLegend()
p2 <- RidgePlot(manno.seurat, features = "ACTB", group.by = "Source")
p1 + p2

3 Seurat与loom的相互转换

还记得上次在单细胞交响乐16-处理大型数据中说到：处理大型数据遇到内存不足时，可以使用这个HDF5ArrayR包（类似的还有 bigmemory, matter），它会将底层数据做成HDF5格式，用硬盘空间来存储数据，必要时再调用一部分数据到内存。loom格式就是处理HDF5使用的

3.1 Seurat转为loom

pbmc.loom <- as.loom(pbmc, filename = "pbmc3k.loom", verbose = FALSE)
pbmc.loom
## Class: loom
## Filename: /__w/1/s/output/pbmc3k.loom
## Access type: H5F_ACC_RDWR
## Attributes: version, chunks, LOOM_SPEC_VERSION, assay, last_modified
## Listing:
##        name    obj_type dataset.dims dataset.type_class
##   col_attrs   H5I_GROUP         <NA>               <NA>
##  col_graphs   H5I_GROUP         <NA>               <NA>
##      layers   H5I_GROUP         <NA>               <NA>
##      matrix H5I_DATASET 2638 x 13714          H5T_FLOAT
##   row_attrs   H5I_GROUP         <NA>               <NA>
##  row_graphs   H5I_GROUP         <NA>               <NA>

# 最后使用完要记得关上loom对象
pbmc.loom$close_all()

3.2 loom转为Seurat

首先读取：用 loomR 的connect

l6.immune <- connect(filename = "../data/l6_r1_immune_cells.loom", mode = "r")
l6.immune
## Class: loom
## Filename: /__w/1/s/data/l6_r1_immune_cells.loom
## Access type: H5F_ACC_RDONLY
## Attributes: CreationDate, last_modified
## Listing:
##        name    obj_type  dataset.dims dataset.type_class
##   col_attrs   H5I_GROUP          <NA>               <NA>
##  col_graphs   H5I_GROUP          <NA>               <NA>
##      layers   H5I_GROUP          <NA>               <NA>
##      matrix H5I_DATASET 14908 x 27998          H5T_FLOAT
##   row_attrs   H5I_GROUP          <NA>               <NA>
##  row_graphs   H5I_GROUP          <NA>               <NA>

然后转换

l6.seurat <- as.Seurat(l6.immune)
VlnPlot(l6.seurat, features = c("Sparc", "Ftl1", "Junb", "Ccl4"), ncol = 2, pt.size = 0.1)

最后处理完，记得关闭loom文件

l6.immune$close_all()

3.3 补充

如果使用Seurat V2，还有一个自带的函数Convert

data("pbmc_small")
pbmc_small
pfile <- Convert(from = pbmc_small, to = "loom", filename = "pbmc_small.loom", 
    display.progress = FALSE)
pfile
## Class: loom
## Filename: /home/paul/Documents/Satija/pbmc_small.loom
## Access type: H5F_ACC_RDWR
## Attributes: version, chunks
## Listing:
##        name    obj_type dataset.dims dataset.type_class
##   col_attrs   H5I_GROUP         <NA>               <NA>
##  col_graphs   H5I_GROUP         <NA>               <NA>
##      layers   H5I_GROUP         <NA>               <NA>
##      matrix H5I_DATASET     80 x 230          H5T_FLOAT
##   row_attrs   H5I_GROUP         <NA>               <NA>
##  row_graphs   H5I_GROUP         <NA>               <NA>

4 Scanpy转Seurat

Seurat有一个函数ReadH5AD可以读取AnnData的H5AD文件

pbmc3k <- ReadH5AD(file = "pbmc3k.h5ad")
# 利用Seurat操作
Idents(pbmc3k) <- "louvain"
p1 <- DimPlot(pbmc3k, label = TRUE)
p2 <- VlnPlot(pbmc3k, features = c("CST3", "NKG7", "PPBP"), combine = FALSE)
wrap_plots(c(list(p1), p2), ncol = 2) & NoLegend()

目前还不能直接将Seurat写成H5AD文件，因此不能之间将Seurat转为Scanpy；但是可以将loom文件作为桥梁实现Seurat转Scanpy，例如Scanpy 有一个函数scanpy.read_loom()

参考：https://scanpy.readthedocs.io/en/stable/api/scanpy.read_loom.html