本文是参考学习 CNS图表复现01—读入csv文件的表达矩阵构建Seurat对象
的学习笔记。可能根据学习情况有所改动。
我从这些CNS文章里面精挑细选了一个非常值得大家花时间跟下去的，就是新鲜出炉的发表在CELL杂志的：Therapy-Induced Evolution of Human Lung Cancer Revealed by Single-Cell RNA Sequencing 。全套代码在：https://github.com/czbiohub/scell_lung_adenocarcinoma
而且是以rmarkdown形式组织的条理清楚，目录如下:
01_Import_data_and_metadata.Rmd
02.1_Create_Seurat_object_neo_osi.Rmd
02_Create_Seurat_object.Rmd
03.1_Subset_and_general_annotations.Rmd
03_Merge_in_NeoOsi.Rmd
03_Subset_and_general_annotations.Rmd
IM01_Subset_cluster_annotate_immune_cells.Rmd
IM02_immune_cell_changes_with_response_to_treatment.Rmd
IM03_Subset_cluster_annotate_MFs-monocytes_LUNG.Rmd
IM04_Subset_cluster_annotate_T-cells_LUNG.Rmd
IM05_Immune_cells_across_pats_with_multiple_biopsies.Rmd
IM06_Combine_Immune_and_nonImmune_annotations.Rmd
NI01_General_annotation_of_nonimmune_cells.Rmd
NI02_epi_subset_and_cluster.Rmd
NI03_inferCNV.Rmd
NI04_Cancer_cells_DEgenes.Rmd
NI05_Annotation_of_Nontumor_epi.Rmd
NI06_mutation_analysis.Rmd
NI07_TH226_cancercell_analysis.Rmd
NI08_Gene_expression_plotting.Rmd
NI09_AT2_sig_compare.Rmd
NI10_TCGA_clinical_outcomes.Rmd
NI14_qpcr_analysis.Rmd
NI15_multiplex_IF_analysis.Rmd
NI16_cancercell_EGFR_ALK.Rmd
NI16_regression_analyses.Rmd
NI17_cancercell_PDsigs.Rmd
V01_various_small_plots.Rmd
我们这个CNS文章是直接把表达矩阵给出来了csv文件，所以直接读取，代码如下：
rm(list=ls())
options(stringsAsFactors = F)
library(Seurat)
# Load data
dir='./'
Sys.time()
raw.data <- read.csv(paste(dir,"Data_input/csv_files/S01_datafinal.csv", sep=""), header=T, row.names = 1)
Sys.time()
dim(raw.data)
raw.data[1:4,1:4]
head(colnames(raw.data))
# Load metadata
metadata <- read.csv(paste(dir,"Data_input/csv_files/S01_metacells.csv", sep=""), row.names=1, header=T)
head(metadata)
# Find ERCC's, compute the percent ERCC, and drop them from the raw data.
erccs <- grep(pattern = "^ERCC-", x = rownames(x = raw.data), value = TRUE)
percent.ercc <- Matrix::colSums(raw.data[erccs, ])/Matrix::colSums(raw.data)
fivenum(percent.ercc)
ercc.index <- grep(pattern = "^ERCC-", x = rownames(x = raw.data), value = FALSE)
raw.data <- raw.data[-ercc.index,]
dim(raw.data)
构建Seurat对象
有了表达矩阵，直接使用 CreateSeuratObject 函数即可，然后慢慢添加这个表达矩阵的一些其它外部属性，全部代码如下：
# Create the Seurat object with all the data (unfiltered)
main_tiss <- CreateSeuratObject(counts = raw.data)
# add rownames to metadta
row.names(metadata) <- metadata$cell_id
# add metadata to Seurat object
main_tiss <- AddMetaData(object = main_tiss, metadata = metadata)
main_tiss <- AddMetaData(object = main_tiss, percent.ercc, col.name = "percent.ercc")
# Head to check
head(main_tiss@meta.data)
# Calculate percent ribosomal genes and add to metadata
ribo.genes <- grep(pattern = "^RP[SL][[:digit:]]", x = rownames(x = main_tiss@assays$RNA@data), value = TRUE)
percent.ribo <- Matrix::colSums(main_tiss@assays$RNA@counts[ribo.genes, ])/Matrix::colSums(main_tiss@assays$RNA@data)
fivenum(percent.ribo)
main_tiss <- AddMetaData(object = main_tiss, metadata = percent.ribo, col.name = "percent.ribo")
main_tiss
# Filter cells so that remaining cells have nGenes >= 500 and nReads >= 50000
main_tiss_filtered <- subset(x=main_tiss, subset = nCount_RNA > 50000 & nFeature_RNA > 500)
main_tiss_filtered
我们得到了main_tiss_filtered这个变量，是一个Seurat对象，就可以follow我们的教程后后续分析流程啦
注：
这个系列教程代码有些bug，对R语言基础不好的同学来说，解决这些bug会比较困难，流程可能走不下来，开始的时候做好心理准备就好。不过看看流程作为了解单细胞分析的基础知识还是挺好的。
