进阶 | 热图 | heatmap | 高通量数据可视化经典图形 | pheatmap | ComplexHeatmap

 

2023年08月22日

热图只显示指定的基因，这个只有ComplexHeatmap能做到，但这个包是真的很复杂！！！

参考：https://jokergoo.github.io/ComplexHeatmap-reference/book/more-examples.html#visualize-cell-heterogeneity-from-single-cell-rnaseq

实例：http://localhost:17435/notebooks/data_center/2023_SOX9_Sci_Adv_PB/PB-log2FC.ipynb

 

颜色选取

require(pals)
pal.bands(coolwarm, parula, ocean.haline, brewer.blues, cubicl, kovesi.rainbow, ocean.phase, brewer.paired(12), stepped, brewer.seqseq2,
          main="Colormap suggestions")

　　

https://cran.r-project.org/web/packages/pals/vignettes/pals_examples.html

 

---

 

高通量数据搞来搞去，都离不开那几图来呈现结果，因为那些图实在是太经典了，高度概括了测序数据的分析结果。

 

热图heatmap：

1. 基因表达矩阵的直观可视化
2. 数据之间的两两比较，相关性分析，相似性分析

 

可视化的对象也非常简单，就是一个表，也叫dataframe或matrix。

 

我最喜欢用pheatmap这个R包，非常好用，也基本涵盖了你想要的所有功能。

 

我的数据，就是两组样本的相关性矩阵，来看相似性。

corM <- cor(as.matrix(human.BP.expr), as.matrix(mouse.BP.expr), method = "spearman")

　　

 

library(pheatmap)
library(ggplot2)

　　

改个label的名字

rownames(corM) <- plyr::mapvalues(rownames(corM), 
                from = c("IMR_ENCC","UE_ENCC","HSCR_5c3","HSCR_20c7","HSCR_10c2","HSCR_1c11","HSCR_17c8","HSCR_23c9","HSCR_6c5"),
                to = c("IMR90","UE02302","HSCR#5","HSCR#20","HSCR#10","HSCR#1","HSCR#17","HSCR#23","HSCR#6"))

　　

colnames(corM) <- plyr::mapvalues(colnames(corM), 
                from = c("E13.5_CNCC"),
                to = c("E13.5 "))

　　

设置顺序

corM <- corM[c("IMR90","UE02302","HSCR#5","HSCR#20","HSCR#10","HSCR#1","HSCR#17","HSCR#23","HSCR#6"),
             c("E9.5","E10.5","E11.5","E12.5","E13.5","E16.5","P19","P21","E13.5 ")]

　　

设置侧边栏的bar的分组和颜色

annotation_row = data.frame(Severity=c("Control","Control","S-HSCR","S-HSCR","S-HSCR","S-HSCR","L-HSCR","L-HSCR","TCA"),
                            row.names=c("IMR90","UE02302","HSCR#5","HSCR#20","HSCR#10","HSCR#1","HSCR#17","HSCR#23","HSCR#6"))

annotation_col = data.frame(row.names=c("E9.5","E10.5","E11.5","E12.5","E13.5","E16.5","P19","P21","E13.5 "),
                            Lineage=c("ENCC","ENCC","ENCC","ENCC","ENCC","ENCC","ENCC","ENCC","CNCC"))

annotation_colors = list(Severity=c(Control="#4DAF4A", `S-HSCR`="#E41A1C", `L-HSCR`="#984EA3", TCA="gray70"),
                         Lineage=c(ENCC="green", CNCC="#eb6566"))

　　

核心画图函数

options(repr.plot.width=7.0, repr.plot.height=6)
p <- pheatmap(corM, cluster_rows = F, cluster_cols = F, display_numbers = T, border_color=NA, 
              number_color="white", fontsize_number = 11, angle_col=315,
              annotation_row=annotation_row, annotation_colors=annotation_colors,
              annotation_col = annotation_col, gaps_col =c(8),
              color = colorRampPalette(rev(hcl.colors(100)))(100))
# brewer.pal(n = 7, name = "RdYlBu")

　　

其中有几个参数要重点说明：

display_numbers：是否显示数值

border_color：是否有边框的线

angle_col：label的旋转，非常方便

gaps_col：自行分组，非常实用的功能，就是拆分，这里不用聚类，随意分

hcl.colors：非常好看的热图的配色

 

保存PDF

pdf("Figures/correlation.enteric.evidence.pdf",width = 7, height = 6)
p
dev.off()

　　

参考：

pheatmap绘制“热图”，你需要的都在这