跟着Nature Communications学作图：R语言ggplot2散点组合误差线展示响应比（Response ratio）

跟着Nature Communications学作图：R语言ggplot2散点组合误差线展示响应比（Response ratio）

论文
Meta-analysis of the impacts of global change factors on soil microbial diversity and functionality

论文里提供了数据和代码，很好的学习素材

这篇论文是公众号的一位读者留言，说这篇论文提供了数据和代码，但是代码比较长，看起来比较吃力。我看了论文中提供的代码，大体上能够看懂，争取抽时间把论文中提供的代码都复现一下。因为论文中的图都对应着提供了作图数据，我们想复现论文中的图。关于用原始数据分析的部分后续有时间在单独介绍。

论文中提供的代码链接
.1038%2Fs41467-020-16881-7/MediaObjects/41467_2020_16881_MOESM8_ESM.txt

今天的推文我们复现论文中的figure1

论文中提供的作图数据如下，excel存储

加载需要用到的R包

library(readxl)
library(tidyverse)
library(latex2exp)
library(ggplot2)

读取数据

metaresult<-read_excel("data/20221129/41467_2020_16881_MOESM9_ESM.xlsx",sheet = 'Fig1')
colnames(metaresult)

首先是第一个小图a
论文中的代码是用RR作为Y轴，GCFs作为X轴，然后再通过coord_flip()函数整体旋转；论文中关于字体上小标是用expression函数实现的，这里我们使用latex2exp这个R包

代码我们参考论文中的代码，但是不完全按照他的写

数据整理和作图代码

data1<-metaresult %>% filter(Variables=="Richness"|Variables=="Shannon")data1$GCFsdata1<-data1 %>% mutate(GCFs=factor(GCFs,levels = c("N_P_K","N_P","N_PPT+","W_eCO2","LUC","N","P","PPT+","PPT-","eCO2","W"))
)data1 %>% colnames()ggplot(data = data1,aes(x=`Weighted means of RR`,y=`GCFs`,shape=Variables))+geom_vline(xintercept=0,linetype = "dashed",size=0.2)+geom_errorbarh(aes(xmin=`Lower confidence intervals`,xmax=`Upper confidence intervals`),position=position_dodge(0.8),height=0.2)+geom_point(position=position_dodge(0.8), size=3, stroke = 0.3,aes(fill=GCFs),show.legend = FALSE)+geom_text(aes(y =`GCFs` , x = `Upper confidence intervals`+0.015, label = `Sample sizes`),position = position_dodge(width = 0.8),vjust = 0.4, hjust=0.4, size = 4, check_overlap = FALSE)+geom_segment(y = 11.6, x = -Inf, yend = 11.6, xend = Inf, colour = "black",size=0.4)+scale_shape_manual(values=c("Richness"=21,"Shannon"=22))+scale_x_continuous(limits=c(-0.17,0.17),breaks = c(-0.16,-0.08,0,0.08,0.16))+scale_y_discrete(breaks=c("N_P_K","N_P","N_PPT+","W_eCO2","LUC","N","P","PPT+","PPT-","eCO2","W"),labels=c(TeX(r"($N \times P \times K$)"),TeX(r"($N \times P$)"),TeX(r"($N \times PPT$+)"),TeX(r"($W \times eCO_2$)"),"LUC","N","P","PPT+","PPT-",TeX(r"($eCO_2$)"),"W"))+labs(x = "Global change factors ", y = "RR of alpha diversity",colour = 'black')+theme(legend.title = element_blank(),legend.position=c(0.2,0.94),legend.key = element_rect(fill = "white",size = 2),legend.key.width = unit(0.5,"lines"),legend.key.height= unit(0.8,"lines"),legend.background = element_blank(),legend.text=element_text(size=6),panel.background = element_rect(fill = 'white', colour = 'white'),axis.title=element_text(size=9),axis.text.y = element_text(colour = 'black', size = 8),axis.text.x = element_text(colour = 'black', size = 8),axis.line = element_line(colour = 'black',size=0.4),axis.line.y = element_blank(),axis.ticks = element_line(colour = 'black',size=0.4),axis.ticks.y = element_blank())

输出结果

小图b

data2<-metaresult %>% filter(Variables=="Beta Diversity")data2$GCFsdata2<-data2 %>% mutate(GCFs=factor(GCFs,levels = c("N_P_K","N_P","N_PPT+","W_eCO2","LUC","N","P","PPT+","PPT-","eCO2","W")))data2 %>% colnames()ggplot(data = data2,aes(x=`Weighted means of RR`,y=`GCFs`))+geom_vline(xintercept=0,linetype = "dashed",size=0.2)+geom_errorbarh(aes(xmin=`Lower confidence intervals`,xmax=`Upper confidence intervals`),height=0.2)+geom_point(size=3, stroke = 0.3,shape=21,aes(fill=GCFs),show.legend = FALSE)+geom_text(aes(y =`GCFs` , x = `Upper confidence intervals`+0.1, label = `Sample sizes`),#position = position_dodge(width = 0.8),vjust = 0.4, hjust=0.4, size = 4, check_overlap = FALSE)+geom_segment(y = 11.6, x = -Inf, yend = 11.6, xend = Inf, colour = "black",size=0.4)+scale_x_continuous(limits=c(-0.6,1.1),breaks = c(-0.5,0,0.5,1))+scale_y_discrete(breaks=c("N_P_K","N_P","N_PPT+","W_eCO2","LUC","N","P","PPT+","PPT-","eCO2","W"),labels=c(TeX(r"($N \times P \times K$)"),TeX(r"($N \times P$)"),TeX(r"($N \times PPT$+)"),TeX(r"($W \times eCO_2$)"),"LUC","N","P","PPT+","PPT-",TeX(r"($eCO_2$)"),"W"))+labs(y = "Global change factors ", x = "RR of alpha diversity",colour = 'black')+theme(legend.title = element_blank(),legend.position=c(0.2,0.9),legend.key = element_rect(fill = "white",size = 2),legend.key.width = unit(0.5,"lines"),legend.key.height= unit(0.8,"lines"),legend.background = element_blank(),legend.text=element_text(size=6),panel.background = element_rect(fill = 'white', colour = 'white'),axis.title=element_text(size=9),axis.text.y = element_text(colour = 'black', size = 8),axis.text.x = element_text(colour = 'black', size = 8),axis.line = element_line(colour = 'black',size=0.4),axis.line.y = element_blank(),axis.ticks = element_line(colour = 'black',size=0.4),axis.ticks.y = element_blank())

小图c

data3<-metaresult %>% filter(Variables=="Community structure")data3$GCFsdata3<-data3 %>% mutate(GCFs=factor(GCFs,levels = c("N_P_K","N_P","N_PPT+","W_eCO2","LUC","N","P","PPT+","PPT-","eCO2","W")))data3 %>% colnames()ggplot(data = data3,aes(x=`Weighted means of RR`,y=`GCFs`))+geom_vline(xintercept=0,linetype = "dashed",size=0.2)+geom_errorbarh(aes(xmin=`Lower confidence intervals`,xmax=`Upper confidence intervals`),height=0.2)+geom_point(size=3, stroke = 0.3,shape=21,aes(fill=GCFs),show.legend = FALSE)+geom_text(aes(y =`GCFs` , x = `Upper confidence intervals`+0.1, label = `Sample sizes`),#position = position_dodge(width = 0.8),vjust = 0.4, hjust=0.4, size = 4, check_overlap = FALSE)+geom_segment(y = 11.6, x = -Inf, yend = 11.6, xend = Inf, colour = "black",size=0.4)+scale_x_continuous(limits=c(-0.6,2.0),breaks = c(-0.5,0,0.5,1,1.5,2.0))+scale_y_discrete(breaks=c("N_P_K","N_P","N_PPT+","W_eCO2","LUC","N","P","PPT+","PPT-","eCO2","W"),labels=c(TeX(r"($N \times P \times K$)"),TeX(r"($N \times P$)"),TeX(r"($N \times PPT$+)"),TeX(r"($W \times eCO_2$)"),"LUC","N","P","PPT+","PPT-",TeX(r"($eCO_2$)"),"W"))+labs(y = "Global change factors ", x = "RR of community structure",colour = 'black')+theme(legend.title = element_blank(),legend.position=c(0.2,0.9),legend.key = element_rect(fill = "white",size = 2),legend.key.width = unit(0.5,"lines"),legend.key.height= unit(0.8,"lines"),legend.background = element_blank(),legend.text=element_text(size=6),panel.background = element_rect(fill = 'white', colour = 'white'),axis.title=element_text(size=9),axis.text.y = element_text(colour = 'black', size = 8),axis.text.x = element_text(colour = 'black', size = 8),axis.line = element_line(colour = 'black',size=0.4),axis.line.y = element_blank(),axis.ticks = element_line(colour = 'black',size=0.4),axis.ticks.y = element_blank())

图b和图c是一样的

最后是拼图
论文中提供的拼图代码是用ggpubr这个R包做的

ggpubr::ggarrange(p1, p2, p3, widths = c(7, 5, 5),ncol = 3, nrow = 1, labels = c("a", "b", "c"), font.label=list(size=12),hjust = 0, vjust = 1)

我自己更习惯使用patchwork这个R包

library(patchwork)p1+p2+theme(axis.text.y = element_blank(),axis.title.y = element_blank())+p3+theme(axis.text.y = element_blank(),axis.title.y = element_blank())+plot_annotation(tag_levels = "a")+plot_layout(widths = c(7, 5, 5))

最终结果

示例数据和代码可以自己到论文中下载，如果需要我推文中的代码和数据可以给公众号推文点赞，点击在看，最后留言获取

查rma()函数找到了这个链接

:~:text=The%20function%20rma()%20(random,compute%20effect%20sizes%20before%20modelling.&text=Random%20effect%20model%20can%20be,%2D%2D%2DFixed%20effect%20model%20cannot.

.html