🤗 Harmony | 超好用的单细胞测序数据合并（3'和5'数据合并）（二）

生信漫卷

发布于 2023-02-24 13:51:14

8130

发布于 2023-02-24 13:51:14

文章被收录于专栏：R语言及实用科研软件

1写在前面

上一期我们介绍了常用的三种合并datasets的方法: 👇

Harmony;
rliger;
Seurat。

本期我们继续介绍其中的harmony包，如何用于3'和5'数据的合并。🤒

2用到的包

rm(list = ls())
library(Seurat)
library(SeuratDisk)
library(SeuratWrappers)
library(patchwork)
library(harmony)
library(rliger)
library(RColorBrewer)
library(tidyverse)
library(reshape2)
library(ggsci)
library(ggstatsplot)

3示例数据

这里我们提供1个3’ PBMC dataset和1个5’ PBMC dataset。🧐

matrix_3p <- Read10X_h5("./3p_pbmc10k_filt.h5",use.names = T)
matrix_5p <- Read10X_h5("./5p_pbmc10k_filt.h5",use.names = T)$`Gene Expression`

srat_3p <- CreateSeuratObject(matrix_3p,project = "pbmc10k_3p")
srat_5p <- CreateSeuratObject(matrix_5p,project = "pbmc10k_5p")
srat_3p
srat_5p

Note! 5' datset中还有一个assay，即VDJ data。🤜

4初步合并

4.1 简单合并

这里我们先用merge将2个数据集简单合并在一起。（这里我们默认做过初步过滤了哈，具体的大家可以看一下上期的教学。）😘

pbmc_harmony <- merge(srat_3p,srat_5p)

4.2 标准操作

我们在这里做一下Normalization，寻找高变基因等标准操作。👀

pbmc_harmony <- NormalizeData(pbmc_harmony, verbose = F)
pbmc_harmony <- FindVariableFeatures(pbmc_harmony, selection.method = "vst", nfeatures = 2000, verbose = F)
pbmc_harmony <- ScaleData(pbmc_harmony, verbose = F)
pbmc_harmony <- RunPCA(pbmc_harmony, npcs = 30, verbose = F)
pbmc_harmony <- RunUMAP(pbmc_harmony, reduction = "pca", dims = 1:30, verbose = F)

5harmony合并数据

5.1 合并前

harmony合并前，PCA明显分离。🥲

DimPlot(pbmc_harmony, reduction = "umap") + 
  scale_color_npg()+
  plot_annotation(title = "10k 3' PBMC and 10k 5' PBMC cells, before integration")

5.2 开始合并

pbmc_harmony <- pbmc_harmony %>% 
  RunHarmony("orig.ident", plot_convergence = T)

5.3 查看信息

harmony_embeddings <- Embeddings(pbmc_harmony, 'harmony')
harmony_embeddings[1:5, 1:5]

5.4 可视化-harmony

p1 <- DimPlot(object = pbmc_harmony, reduction = "harmony", 
              pt.size = .1, group.by = "orig.ident") + 
  scale_color_npg()+
  NoLegend()

p2 <- VlnPlot(object = pbmc_harmony, features = "harmony_1", 
              group.by = "orig.ident", pt.size = .1) + 
  scale_color_npg()+
  NoLegend()

p1 + p2

5.5 可视化-UMAP

harmony合并后，UMAP几乎重叠，但效果似乎没有Seurat包好。🤒

pbmc_harmony <- pbmc_harmony %>% 
  RunUMAP(reduction = "harmony", dims = 1:30, verbose = F)

pbmc_harmony <- SetIdent(pbmc_harmony,value = "orig.ident")
p1 <- DimPlot(pbmc_harmony,reduction = "umap") + 
  scale_color_npg()+
  plot_annotation(title = "10k 3' PBMC and 10k 5' PBMC cells, after integration (Harmony)")

p2 <- DimPlot(pbmc_harmony, reduction = "umap", 
              group.by = "orig.ident", pt.size = .1, 
              split.by = 'orig.ident') + 
  scale_color_npg()+
  NoLegend()

p1 + p2

6降维与聚类

6.1 寻找clusters

pbmc_harmony <- pbmc_harmony %>% 
  FindNeighbors(reduction = "harmony", k.param = 10, dims = 1:30) %>% 
  FindClusters() %>% 
  identity()

6.2 聚类可视化

pbmc_harmony <- SetIdent(pbmc_harmony,value = "seurat_clusters")

ncluster <- length(unique(pbmc_harmony[[]]$seurat_clusters))

mycol <- colorRampPalette(brewer.pal(8, "Set2"))(ncluster)

DimPlot(pbmc_harmony,label = T,
        cols = mycol, repel = T) + 
  NoLegend()

6.3 具体查看及可视化

我们看下各个clusters在两个datasets各有多少细胞。

count_table <- table(pbmc_harmony@meta.data$seurat_clusters, pbmc_harmony@meta.data$orig.ident)
count_table

#### 可视化
count_table %>% 
  as.data.frame() %>% 
  ggbarstats(x = Var2, 
             y = Var1,
             counts = Freq)+
  scale_fill_npg()