GEO生物信息学与算法R语言训练

在火山图上标记基因

2019-07-17  本文已影响57人  小洁忘了怎么分身

要玩图,离不开哈德雷大神的《R数据科学》,第1章和21章是专门讲图的,我写过对应的笔记:
https://www.jianshu.com/p/4a154f6f0de7
https://www.jianshu.com/p/bf0f12246865

关于火山图加标签的需求,这里有几种方法来实现。

示例数据

方法一的示例数据是data.Rdata,方法二三的示例数据是test.Rdata。我将数据打包放在了“生信星球”公众号后台,回复“火山图”即可获得。你解压后双击文件夹里的volcano.Rproj,复制粘贴运行本文代码即可。


方法一:利用空字符串“”

关于空字符串我曾写过一篇文章来讲他:https://www.jianshu.com/p/aef98f3fc7d8

这种方法的参照是帮助文档里的一段代码:
(先准备好包)

if(!require(ggplot2)) install.packages("ggplot2")
if(!require(ggrepel)) install.packages("ggrepel")
library(ggplot2)
library(ggrepel)

下面代码来源于geom_text_repel的帮助文档

p <- ggplot(mtcars,
            aes(wt, mpg, label = rownames(mtcars), colour = factor(cyl))) +
  geom_point()
# Hide some of the labels, but repel from all data points
mtcars$label <- rownames(mtcars)
mtcars$label[1:15] <- ""
p + geom_text_repel(data = mtcars, aes(wt, mpg, label = label))

做出的图是这样:



可以看到,一部分点有标签, 一部分没有,思路就是把不要标签的部分变成空字符串“”。
那么参考这个思路为火山图加标签:
(美图预警)
先把图画出来:

load("data.Rdata")
head(data)
#       symbol     p.value        FC change 
#1            PCMTD2 1.53544e-11 1.3548360 Stable      
#2                KIAA0087 6.71382e-13 0.7314603 Stable      
#3                 AFAP1L1 4.24611e-12 0.6284560 Stable      
#4                  CHMP1A 3.76821e-09 1.6035994 Stable      
#5                  TRERF1 1.80652e-08 0.6875469 Stable      
#6                     C8B 7.88047e-04 1.2374303 Stable      
data$change = ifelse(data$p.value < 0.000001 & abs(log2(data$FC)) >= 1, 
                        ifelse(log2(data$FC)> 1 ,'Up','Down'),
                        'Stable')

p <- ggplot(data = data, 
         aes(x = log2(data$FC), 
             y = -log10(data$p.value), 
             colour=change,
             label = data$symbol)) +
  geom_point(alpha=0.4, size=3.5) +
  scale_color_manual(values=c("blue", "grey","red"))+
  xlim(c(-4.5, 4.5)) +
  geom_vline(xintercept=c(-1,1),lty=4,col="black",lwd=0.8) +
  geom_hline(yintercept = -log10(0.000001),lty=4,col="black",lwd=0.8) +
  labs(x="log2(fold change)",
       y="-log10 (p-value)",
       title="Differential metabolites")  +
  theme_bw()+
  theme(plot.title = element_text(hjust = 0.5), 
        legend.position="right", 
        legend.title = element_blank())
p

然后是加标签,重点就在这里:

data$label=ifelse(data$p.value < 0.000001 & abs(log2(data$FC)) >= 1,data$symbol,"")
p+geom_text_repel(data = data, aes(x = log2(data$FC), 
                                   y = -log10(data$p.value), 
                                   label = label),
                      size = 3,box.padding = unit(0.5, "lines"),
                      point.padding = unit(0.8, "lines"), 
                      segment.color = "black", 
                      show.legend = FALSE)

但是我发现,这个只是适用于数据量比较小的时候,一般来说火山图数以万计的行,用这个方法容易失败。下午尝试了几次大的数据,结果Rstudio无一例外的嘎嘣了。

方法二:看R数据科学

以下代码出自R数据科学笔记第21章:

best_in_class <- mpg %>%
  group_by(class) %>%
  filter(row_number(desc(hwy)) == 1)

ggplot(mpg, aes(displ, hwy)) +
  geom_point(aes(color = class)) +
  geom_point(size = 3, shape = 1, data = best_in_class) +
  ggrepel::geom_label_repel(
    aes(label = model),
    data = best_in_class
  )

这个方法适用于较大的数据。
先把图画出:

load("test.Rdata")
p <- ggplot(data = test, 
            aes(x = logFC, 
                y = `-log10(P.value)`)) +
  geom_point(alpha=0.4, size=3.5, 
             aes(color=change)) +
  scale_color_manual(values=c("blue", "grey","red"))+
  geom_vline(xintercept=c(-1,1),lty=4,col="black",lwd=0.8) +
  geom_hline(yintercept = -log10(0.01),lty=4,col="black",lwd=0.8) +
  theme_bw()
p

然后加标签,可以自定义数据框的阈值来调整标签的数量:


for_label <- test %>% 
  filter(abs(logFC) >4& `-log10(P.value)`> -log10(0.000001))

p +
  geom_point(size = 3, shape = 1, data = for_label) +
  ggrepel::geom_label_repel(
    aes(label = symbol),
    data = for_label,
    color="black"
  )

其实原理就是叠加了一个新数据框的两个图层,一个空心点图,一个geom_label_repel。

3.方法三:ggpubr的函数有现成的参数

这个函数叫ggscatter,还是用刚才的test数据来做
由于ggpubr写纵坐标时直接写-log10(P.value)不识别,我采取了迂回策略,改列名,完事再在图上改纵轴标签。

load("test.Rdata")
if(!require(ggpubr))install.packages("ggplubr")
library(ggpubr)
colnames(test)[4] <- "v"
ggscatter(test, 
          x = "logFC", 
          y ="v",
          ylab="-log10(P.value)",
          size=0.5,
          color = "change",
          palette = c("#00AFBB", "#999999", "#FC4E07") 
          )

然后加标签,是县城的参数“label.select”。接受的参数数据结构应该是向量。
可以手动选一二三四个感兴趣的基因:

ggscatter(test, 
          x = "logFC", 
          y = "v", 
          ylab="-log10(P.value)",
          color = "change",
          size = 0.5,
          label = "symbol", 
          repel = T,
          palette = c("#00AFBB", "#999999", "#FC4E07") ,
          #label.select = dat$symbol[1:30] ,
          label.select = c("CD36", "DUSP6", "DCT", "SPRY2", "MOXD1", "ETV4" )
          )

也可以用向量取子集的方法来取很多个,比如差异基因前30个:

ggscatter(test, 
          x = "logFC", 
          y = "v", 
          ylab="-log10(P.value)",
          color = "change",
          size = 0.5,
          label = "symbol", 
          repel = T,
          palette = c("#00AFBB", "#999999", "#FC4E07") ,
          label.select = test$symbol[1:30]
          )
上一篇下一篇

猜你喜欢

热点阅读