CytoTRACE2可视化进阶(修改坐标维持umap图前后一致)

原创

凑齐六个字吧

发布于 2024-09-19 00:28:15

6734

文章被收录于专栏：单细胞单细胞

上一期推文已经简单说明了CytoTRACE2的分析流程、注意事项和可视化。

CytoTRACE2单细胞分化潜力预测工具学习：https://mp.weixin.qq.com/s/inF9iGy2X9D2CZLzQBtRdw

有小伙伴观察到使用CytoTRACE2之后的图形会与原始的umap图有些许的不同，虽然在实际应用的时候这种情况是无伤大雅的。

但既然提出了问题，笔者就尝试着去解决一下~

步骤流程

1、导入

rm(list=ls())
library(tidyverse)
library(CytoTRACE2) 
library(Seurat)
library(paletteer)
library(BiocParallel)
register(MulticoreParam(workers = 4, progressbar = TRUE))

load("scRNA.Rdata")
sub_dat <- subset(scRNA, subset = celltype %in% c("CD8+ T-cells", "CD4+ T-cells"))
DimPlot(sub_dat,label = T)

2、提取数据/运行CytoTRACE2

# 提取表达矩阵信息
expression_data <- GetAssayData(sub_dat,layer = "counts")

# 运行CytoTRACE2
cytotrace2_result <- cytotrace2(expression_data,species = 'human')

# 提取注释信息
annotation <- data.frame(phenotype = sub_dat@meta.data$celltype) %>% 
  set_rownames(., colnames(sub_dat))

# 使用plotData函数生成预测和表型关联图
plots <- plotData(cytotrace2_result = cytotrace2_result, 
                  annotation = annotation,
                  expression_data = expression_data
)

3、坐标修改

umap_raw <- as.data.frame(sub_dat@reductions$umap@cell.embeddings) # 为后面修改坐标有用
# 创建一个包含所有需要更新的 plot 名称的向量
plot_names <- c("CytoTRACE2_UMAP", "CytoTRACE2_Potency_UMAP", 
                "CytoTRACE2_Relative_UMAP", "Phenotype_UMAP" 
)
# 循环遍历每个plot并更新坐标
for (plot_name in plot_names) {
  if (!is.null(plots[[plot_name]][[1]]$data)) {
    plots[[plot_name]][[1]]$data$umap_1 <- umap_raw$umap_1
    plots[[plot_name]][[1]]$data$umap_2 <- umap_raw$umap_2
  }
}

# 了解数据类型
class(plots$CytoTRACE2_UMAP[[1]]) 
# [1] "gg"     "ggplot"
class(plots$CytoTRACE2_Potency_UMAP[[1]]) 
# [1] "gg"     "ggplot"
class(plots$CytoTRACE2_Relative_UMAP[[1]]) 
# [1] "gg"     "ggplot"
class(plots$Phenotype_UMAP[[1]])
# [1] "gg"     "ggplot"

# 确认umap的数据范围
x_limits <- range(umap_raw$umap_1, na.rm = TRUE)
y_limits <- range(umap_raw$umap_2, na.rm = TRUE)

# 导出图片
plots$CytoTRACE2_UMAP[[1]] <- plots$CytoTRACE2_UMAP[[1]] + 
  scale_x_continuous(limits = x_limits) + 
  scale_y_continuous(limits = y_limits)
plots$CytoTRACE2_UMAP

plots$CytoTRACE2_Potency_UMAP[[1]] <- plots$CytoTRACE2_Potency_UMAP[[1]] + 
  coord_cartesian(xlim = x_limits, ylim = y_limits)
plots$CytoTRACE2_Potency_UMAP

plots$CytoTRACE2_Relative_UMAP[[1]] <- plots$CytoTRACE2_Relative_UMAP[[1]] + 
  coord_cartesian(xlim = x_limits, ylim = y_limits)
plots$CytoTRACE2_Relative_UMAP

plots$Phenotype_UMAP[[1]] <- plots$Phenotype_UMAP[[1]] + 
  coord_cartesian(xlim = x_limits, ylim = y_limits)
plots$Phenotype_UMAP

library(patchwork)
# 拼接为2x2布局
combined_plot <- (p1[[1]] | p2[[1]]) / (p3[[1]] | p4[[1]])
# 显示拼接后的图形
combined_plot
ggsave("combined_plot.png", plot = combined_plot, width = 10, height = 10)