数据介绍：

来自谷歌翻译：

我们对 44 名患者中 58 例肺腺癌衍生的 208,506 个细胞进行了单细胞 RNA 测序（scRNA-seq），涵盖原发肿瘤、淋巴结和脑转移及胸腔积液，此外还包括正常肺组织和淋巴结。丰富的单细胞谱描绘了肺腺癌进展的复杂细胞图谱，包括周围肿瘤微环境中的癌症、间质和免疫细胞。

数据下载地址：https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE131907

数据下载：

wget -c "https://ftp.ncbi.nlm.nih.gov/geo/series/GSE131nnn/GSE131907/suppl/GSE131907%5FLung%5FCancer%5Fraw%5FUMI%5Fmatrix.txt.gz" -O GSE131907_Lung_Cancer_raw_UMI_matrix.txt.gz
wget -c "https://ftp.ncbi.nlm.nih.gov/geo/series/GSE131nnn/GSE131907/suppl/GSE131907%5FLung%5FCancer%5Fcell%5Fannotation.txt.gz" -O GSE131907_Lung_Cancer_cell_annotation.txt.gz

样本合并的，需要按样本分开方便后续度入质控，这里写了R代码：

#!/usr/bin/env Rscript
# ===============================
# 参数
# ===============================
anno_dir <- "./"   # 存放 LUNG_N01.txt 等文件的目录
umi_file <- "GSE131907_Lung_Cancer_raw_UMI_matrix.txt.gz"
out_dir  <- "./split"
dir.create(out_dir, showWarnings = FALSE)
# ===============================
# 读取 UMI 矩阵
# ===============================
message("Reading UMI matrix...")
umi <- read.table(
  gzfile(umi_file),
  header = TRUE,
  sep = "\t",
  check.names = FALSE,
  stringsAsFactors = FALSE
)
gene_col <- umi[, 1, drop = FALSE]
umi_mat  <- umi[, -1, drop = FALSE]
message("UMI matrix dimensions:")
print(dim(umi_mat))
# ===============================
# 处理每个 Sample
# ===============================
anno_files <- list.files(anno_dir, pattern = "\\.txt$", full.names = TRUE)
for (f in anno_files) {
  sample_name <- sub("\\.txt$", "", basename(f))
  message("Processing ", sample_name)
  # 读取 annotation
  anno <- read.table(
    f,
    header = TRUE,
    sep = "\t",
    stringsAsFactors = FALSE
  )
  idx <- anno$Index
  idx <- intersect(idx, colnames(umi_mat))
  if (length(idx) == 0) {
    warning("No matching cells for ", sample_name)
    next
  }
  sub_mat <- umi_mat[, idx, drop = FALSE]
  out <- cbind(gene_col, sub_mat)
  out_file <- file.path(out_dir, paste0(sample_name, "_UMI.txt.gz"))
  write.table(
    out,
    gzfile(out_file),
    sep = "\t",
    quote = FALSE,
    row.names = FALSE
  )
  message("  Saved: ", out_file,
          " (", ncol(sub_mat), " cells)")
}
message("Done.")

细胞注释文件也分开：

zcat GSE131907_Lung_Cancer_cell_annotation.txt.gz | \
awk -F'\t' '
NR==1 {
  header=$0
  next
}
{
  out=$3 ".txt"
  if (!(out in seen)) {
    print header > out
    seen[out]=1
  }
  print >> out
}'

数据分析：

样本准备mata文件：meta.tsv

这次的数据是​ h5 格式的也可以直接读入：​

cat ~/LUAD/data/meta.tsv | sed '1d' | \
parallel -j 10 --colsep '\t' '
Rscript $scripts/seurat_sc_qc.r \
  --count ~/LUAD/data/split/{1}_UMI.txt.gz \
  -p {1} --project {1} \
  --nUMI.min 100 \
  --nUMI.max 150000 \
  --nGene.min 200 \
  --nGene.max 10000 \
  --mito.gene.pattern "^MT.*-" \
  --percent_mito 20 \
  --log10GenesPerUMI 0.8 \
  --metadata ~/LUAD/data/{1}.txt \
  --metadata.col.name Patient_id Tissue Histology Smoking Pathology EGFR_MUT EGFR_Type Stages Stages_Status \
  --metadata.value {2} {3} {4} {5} {6} {7} {8} {9} {10}
'
#细胞周期和双细胞去除
cat ~/LUAD/data/meta.tsv|sed '1d'|while read Sample Patient_id Tissue Histology Smoking Pathology EGFR_MUT EGFR_Type Stages Stages_Status;do

Rscript $scripts/seurat_sc_cluster.r --rds $Sample.CellCycleScoring.qs  \
 --resolution 0.5 -d 30 \
 -p $Sample   -o $Sample --cpu 20


## 如果是 SCT 标准化需要加参数：--sct
## 如果要去除双细胞增加参数：--removeDoubletCells

Rscript $scripts/DoubletFinder.r -i $Sample/$Sample.qs \
    -p  $Sample   --annotations seurat_clusters --removeDoubletCells
done

#合并样本，
Rscript $scripts/merge_seurat_obj.r -i .doubletFinder.qs   -p all.sample.merged
# 分群聚类

Rscript $scripts/seurat_sc_cluster.r  --cpu 10  --rds all.sample.merged.qs  \
 --integrate.method harmony --batch.id Sample   \
 --resolution 0.2 -d 50 \
 -p luad.harmony  -o luad.harmony 

分析结果：