Meta-analysis code

#forest.meta(m, layout = "RevMan5")
#metainf(m,pooled = "random")

install.packages('metafor')
install.packages('openxlsx')
library(graphics)
library(meta)
library(xlsxjars)
library(xlsx)
library(ggplot2)



#N
x<-paste0("N","Stage.pdf")          #换           
y<-paste0("N","funnel.pdf")          #换   
metadat <- read.xlsx("./meta_R.xlsx", sheetIndex =2) #换   

metabin(a,b,c,d,data = metadat,sm="OR", random = F ,fixed=TRUE)
m<-metabin(a,b,c,d,data = metadat,sm="OR", random = T ,fixed=F,studlab = paste(Author,Year))    
#上面可以加comb.fixed 及comb.random 为效应模型的选择(“固定"，"随机")

pdf(x,width = 7,height = 3)
forest.meta(m,random = T, fixed   = F,xlab = "Lymph node metastasis",
            label.left  = "LOX-",label.right = "LOX+",col.square = "#4682B4",
            col.diamond = "red", col.diamond.lines = "black", 
            layout = "JAMA") #换 #B4464B

dev.off()
pdf(y)
funnel(m,xlab = "Lymph node metastasis")
metabias(m,k.min=4)
dev.off()

#prediction = T,,fixed   = T#4682B4
metainf(m,pooled = "random")
metainf(m,pooled = "fixed")

pdf("N_sens.pdf",width = 7,height = 3)
pic<-forest(metainf(m,pooled = "fixed"),comb.random=F)
pic_N_sens<-forest(metainf(m,pooled = "random"),comb.random=T,xlab = "Lymph node metastasis",
                   label.left  = "LOX-",label.right = "LOX+",col.square = "#4682B4",
                   col.diamond = "red", col.diamond.lines = "black",
                   layout = "JAMA")
dev.off()


#T
x<-paste0("T","Forest.pdf")          #换           
y<-paste0("T","funnel.pdf")          #换   
metadat <- read.xlsx("./meta_R.xlsx", sheetIndex =1) #换   

metabin(a,b,c,d,data = metadat,sm="OR", random = T ,fixed=F)
m<-metabin(a,b,c,d,data = metadat,sm="OR",random = T ,fixed=F,studlab = paste(Author,Year))    
#上面可以加comb.fixed 及comb.random 为效应模型的选择(“固定"，"随机")

pdf(x,width = 7,height = 3)
forest.meta(m,random = T,fixed   =  F, col.square = "#4682B4",
            xlab = "Tumor invasion depth",
            label.left  = "LOX-",label.right = "LOX+",
            col.diamond = "#B4464B", col.diamond.lines = "black", 
            layout = "JAMA")   #换

dev.off()
pdf(y)
funnel(m,xlab ="Tumor invasion depth" )
metabias(m,k.min=4)
dev.off()
?forest.meta
metabias(m,k.min=4)



metainf(m,pooled = "random")
metainf(m,pooled = "fixed")

pdf("T_sens.pdf",width = 7,height = 3)
pic<-forest(metainf(m,pooled = "fixed"),comb.random=F)
pic_T_sens<-forest(metainf(m,pooled = "random"),comb.random=T, col.square = "#4682B4",
                   xlab = "Tumor invasion depth",
                   label.left  = "LOX-",label.right = "LOX+",
                   col.diamond = "#B4464B", col.diamond.lines = "black", 
                   layout = "JAMA")
dev.off()






#M
x<-paste0("M","Forest.pdf")          #换           
y<-paste0("M","funnel.pdf")          #换   
metadat <- read.xlsx("./meta_R.xlsx", sheetIndex =3) #换   

metabin(a,b,c,d,data = metadat,sm="OR", random = T ,fixed=TRUE)
m<-metabin(a,b,c,d,data = metadat,sm="OR",random = T ,fixed=F,studlab = paste(Author,Year))    
#上面可以加comb.fixed 及comb.random 为效应模型的选择(“固定"，"随机")

pdf(x,width = 7,height = 3)
forest.meta(m,random = T,fixed   =  F, col.square = "#4682B4",
            xlab = "Tumor metastasis",
            label.left  = "LOX-",label.right = "LOX+",
            col.diamond = "#B4464B", col.diamond.lines = "black", 
            layout = "JAMA")
dev.off()
pdf(y)
funnel(m,xlab = "Tumor metastasis")
metabias(m,k.min=4)
dev.off()


metainf(m,pooled = "random")
metainf(m,pooled = "fixed")

pdf("M_sens.pdf",width = 7,height = 3)
pic<-forest(metainf(m,pooled = "fixed"),comb.random=F)
pic_M_sens<-forest(metainf(m,pooled = "random"),comb.random=T, col.square = "#4682B4",
                   xlab = "Tumor metastasis",
                   label.left  = "LOX-",label.right = "LOX+",
                   col.diamond = "#B4464B", col.diamond.lines = "black", 
                   layout = "JAMA")
dev.off()






#Differentiation

x<-paste0("D","Forest.pdf")          #换           
y<-paste0("D","funnel.pdf")          #换   
metadat <- read.xlsx("./meta_R.xlsx", sheetIndex =6) #换   

metabin(a,b,c,d,data = metadat,sm="OR", random = T ,fixed=TRUE)
m<-metabin(a,b,c,d,data = metadat,sm="OR",random = F ,fixed=TRUE,studlab = paste(Author,Year))    
#上面可以加comb.fixed 及comb.random 为效应模型的选择(“固定"，"随机")

pdf(x,width = 7,height = 3)
forest.meta(m,random = F,fixed   =  T, col.square = "#4682B4",
            xlab = "Differentiation",
            label.left  = "LOX-",label.right = "LOX+",
            col.diamond = "#B4464B", col.diamond.lines = "black", 
            layout = "JAMA")
dev.off()
pdf(y)
funnel(m,xlab = "Differentiation")
metabias(m,k.min=4)
dev.off()



#敏感性分析
metainf(m,pooled = "random")
metainf(m,pooled = "fixed")

pdf("D_sens.pdf",width = 7,height = 3)
pic_D_sens<-forest(metainf(m,pooled = "fixed"),comb.random=F, col.square = "#4682B4",
                   xlab = "Differentiation",
                   label.left  = "LOX-",label.right = "LOX+",
                   col.diamond = "#B4464B", col.diamond.lines = "black", 
                   layout = "JAMA")
pic_D_sens<-forest(metainf(m,pooled = "random"),comb.random=T)
dev.off()




#Gender
x<-paste0("Gender","Forest.pdf")          #换           
y<-paste0("Gender","funnel.pdf")          #换   
metadat <- read.xlsx("./meta_R.xlsx", sheetIndex =5) #换   

metabin(a,b,c,d,data = metadat,sm="OR", random = T ,fixed=TRUE)
m<-metabin(a,b,c,d,data = metadat,sm="OR",random = T ,fixed=F,studlab = paste(Author,Year))    
#上面可以加comb.fixed 及comb.random 为效应模型的选择(“固定"，"随机")
pdf(x,width = 7,height = 3)
forest.meta(m,random = T,fixed   =  F, col.square = "#4682B4",
            xlab = "Gender",
            label.left  = "LOX-",label.right = "LOX+",
            col.diamond = "#B4464B", col.diamond.lines = "black", 
            layout = "JAMA")
dev.off()
pdf(y)
funnel(m,xlab = "Gender")
metabias(m,k.min=4)
dev.off()
trimfill(m,fixed = T)   #减补法之后

#敏感性分析
metainf(m,pooled = "random")
metainf(m,pooled = "fixed")

pdf("G_sens.pdf",width = 7,height = 3)
pic_D_sens<-forest(metainf(m,pooled = "fixed"),comb.random=F)
pic_G_sens<-forest(metainf(m,pooled = "random"),comb.random=T, col.square = "#4682B4",
                   xlab = "Gender",
                   label.left  = "LOX-",label.right = "LOX+",
                   col.diamond = "#B4464B", col.diamond.lines = "black", 
                   layout = "JAMA")
dev.off()






#Lauren parting

x<-paste0("Lauren classification","Forest.pdf")          #换           
y<-paste0("Lauren classification","funnel.pdf")          #换   
metadat <- read.xlsx("./meta_R.xlsx", sheetIndex =4) #换   

metabin(a,b,c,d,data = metadat,sm="OR", random = T ,fixed=TRUE)
m<-metabin(a,b,c,d,data = metadat,sm="OR",random = F ,fixed=TRUE,studlab = paste(Author,Year))    
#上面可以加comb.fixed 及comb.random 为效应模型的选择(“固定"，"随机")
pdf(x,width = 7,height = 3)
forest.meta(m,random = F,fixed   =  T, col.square = "#4682B4",
            xlab = "Lauren classification",
            label.left  = "LOX-",label.right = "LOX+",
            col.diamond = "#B4464B", col.diamond.lines = "black", 
            layout = "JAMA")
dev.off()
pdf(y)
funnel(m ,xlab = "Lauren classification")
metabias(m,k.min=4)
dev.off()
trimfill(m,fixed = T)
#敏感性分析
metainf(m,pooled = "random")
metainf(m,pooled = "fixed")

pdf("L_sens.pdf",width = 7,height = 3)
pic_L_sens<-forest(metainf(m,pooled = "fixed"),comb.random=F, col.square = "#4682B4",
                   xlab = "Lauren classification",
                   label.left  = "LOX-",label.right = "LOX+",
                   col.diamond = "#B4464B", col.diamond.lines = "black", 
                   layout = "JAMA")
pic_G_sens<-forest(metainf(m,pooled = "random"),comb.random=T)
dev.off()








#1-year OS
x<-paste0("1-year OS","Forest.pdf")          #换           
y<-paste0("1-year OS","funnel.pdf")          #换   
metadat <- read.xlsx("./meta_R.xlsx", sheetIndex =7) #换   

metabin(a,b,c,d,data = metadat,sm="OR", random = T ,fixed=TRUE)
m<-metabin(a,b,c,d,data = metadat,sm="OR",random = T ,fixed=F,studlab = paste(Author,Year))    
#上面可以加comb.fixed 及comb.random 为效应模型的选择(“固定"，"随机")
pdf(x,width = 7,height = 3)
forest.meta(m,random = T,fixed   =  F, col.square = "#4682B4",
            xlab = "1-year OS",
            label.left  = "LOX-",label.right = "LOX+",
            col.diamond = "#B4464B", col.diamond.lines = "black", 
            layout = "JAMA")
dev.off()
pdf(y)
funnel(m ,xlab = "1-year OS")
metabias(m1,k.min=4)
dev.off()
m1<-trimfill(m, random  = TRUE)


#敏感性分析
metainf(m,pooled = "random")
metainf(m,pooled = "fixed")

pdf("1OS_sens.pdf",width = 7,height = 3)
pic_L_sens<-forest(metainf(m,pooled = "fixed"),comb.random=F)
pic_1OS_sens<-forest(metainf(m,pooled = "random"),comb.random=T, col.square = "#4682B4",
                     xlab = "1-year OS",
                     label.left  = "LOX-",label.right = "LOX+",
                     col.diamond = "#B4464B", col.diamond.lines = "black", 
                     layout = "JAMA")
dev.off()






#3-year OS

x<-paste0("3-year OS","Forest.pdf")          #换           
y<-paste0("3-year OS","funnel.pdf")          #换   
metadat <- read.xlsx("./meta_R.xlsx", sheetIndex =8) #换   

metabin(a,b,c,d,data = metadat,sm="OR", random = T ,fixed=TRUE)
m<-metabin(a,b,c,d,data = metadat,sm="OR",random = T ,fixed=F,studlab = paste(Author,Year))    
#上面可以加comb.fixed 及comb.random 为效应模型的选择(“固定"，"随机")
pdf(x,width = 7,height = 3)
forest.meta(m,random = T,fixed   =  F, col.square = "#4682B4",
            xlab = "3-year OS",
            label.left  = "LOX-",label.right = "LOX+",
            col.diamond = "#B4464B", col.diamond.lines = "black", 
            layout = "JAMA")
dev.off()
pdf(y)
funnel(m,xlab = "3-year OS")
metabias(m,k.min=4)
dev.off()


#敏感性分析
metainf(m,pooled = "random")
metainf(m,pooled = "fixed")

pdf("3OS_sens.pdf",width = 7,height = 3)
pic_L_sens<-forest(metainf(m,pooled = "fixed"),comb.random=F)
pic_3OS_sens<-forest(metainf(m,pooled = "random"),comb.random=T, col.square = "#4682B4",
                     xlab = "3-year OS",
                     label.left  = "LOX-",label.right = "LOX+",
                     col.diamond = "#B4464B", col.diamond.lines = "black", 
                     layout = "JAMA")
dev.off()




#5-year OS

x<-paste0("5-year OS","Forest.pdf")          #换           
y<-paste0("5-year OS","funnel.pdf")          #换   
metadat <- read.xlsx("./meta_R.xlsx", sheetIndex =9) #换   

metabin(a,b,c,d,data = metadat,sm="OR", random = T ,fixed=TRUE)
m<-metabin(a,b,c,d,data = metadat,sm="OR",random = T ,fixed=F,studlab = paste(Author,Year))    
#上面可以加comb.fixed 及comb.random 为效应模型的选择(“固定"，"随机")
pdf(x,width = 7,height = 3)
forest.meta(m,random = T,fixed   =  F, col.square = "#4682B4",
            xlab = "5-year OS",
            label.left  = "LOX-",label.right = "LOX+",
            col.diamond = "#B4464B", col.diamond.lines = "black", 
            layout = "JAMA")
dev.off()
pdf(y)
funnel(m, xlab = "5-year OS")
metabias(m,k.min=4)
dev.off()

#敏感性分析
metainf(m,pooled = "random")
metainf(m,pooled = "fixed")

pdf("5OS_sens.pdf",width = 7,height = 3)
pic_L_sens<-forest(metainf(m,pooled = "fixed"),comb.random=F)
pic_5OS_sens<-forest(metainf(m,pooled = "random"),comb.random=T, col.square = "#4682B4",
                     xlab = "5-year OS",
                     label.left  = "LOX-",label.right = "LOX+",
                     col.diamond = "#B4464B", col.diamond.lines = "black", 
                     layout = "JAMA")
dev.off()

install.packages("cowplot")
library(cowplot)
library(tidyverse)
library(ggpubr)
par()
forest(metainf(m,pooled = "random"),comb.random=T, col.square = "#4682B4",
       xlab = "5-year OS",
       label.left  = "LOX-",label.right = "LOX+",
       col.diamond = "#B4464B", col.diamond.lines = "black", 
       layout = "JAMA")








#age


x<-paste0("age","Forest.pdf")          #换           
y<-paste0("age","funnel.pdf")          #换   
metadat <- read.xlsx("./meta_R.xlsx", sheetIndex =10) #换   

metabin(a,b,c,d,data = metadat,sm="OR", random = T ,fixed=TRUE)
m<-metabin(a,b,c,d,data = metadat,sm="OR",studlab = paste(Author,Year))    
#上面可以加comb.fixed 及comb.random 为效应模型的选择(“固定"，"随机")
pdf(x,width = 7,height = 3)
forest.meta(m,random = T,fixed   =  F, col.square = "#4682B4",
            xlab = "age",
            label.left  = "LOX-",label.right = "LOX+",
            col.diamond = "#B4464B", col.diamond.lines = "black", 
            layout = "JAMA")
dev.off()
pdf(y)
funnel(m, xlab = "5-year OS")
metabias(m,k.min=4)
dev.off()


library(rJava)
library(xlsxjars)
library(xlsx)

for ( j in 1:9) {
  
metadat <- read.xlsx("./meta_R.xlsx", sheetIndex =j) 
for (i in 1:nrow(metadat)) {
  
metadat$Conbine[i]<-paste0(metadat[i,1],metadat[i,4],"/",metadat[i,6])
}

write.xlsx(metadat, file="file1.xlsx", sheetName=as.character(j), append=TRUE, row.names=FALSE) 

}



##########合并HR，需要HR，95%置信区间上下，三个数据############
test1 <- read.xlsx("./HR.xlsx", sheetIndex =1) 
test1<-test1[1:5,2:6]
info<-test1
info$lghr <- log(info$hr)
lglci <- log(info$lci)
lguci <- log(info$uci)
info$selghr <- (lguci-lglci)/(2*1.96)
library(meta)
pfs <- metagen(TE = lghr,seTE = selghr,studlab = paste(info$Author,info$Year),data = info,sm = 'HR')



pdf("OS.pdf",width = 7,height = 3)
forest(pfs,random = T,fixed   =  F, col.square = "#4682B4",
       xlab = "OS",
       label.left  = "LOX-",label.right = "LOX+",
       col.diamond = "#B4464B", col.diamond.lines = "black", 
       layout = "JAMA")
dev.off()



#敏感性分析
pdf("OS_sen.pdf",width = 7,height = 3)
forest(metainf(pfs,pooled = "random"),comb.random=T, col.square = "#4682B4",
                     xlab = "OS",
                     label.left  = "LOX-",label.right = "LOX+",
                     col.diamond = "#B4464B", col.diamond.lines = "black", 
                     layout = "JAMA")
dev.off()
#漏斗图
pdf("OS_funnel.pdf")
funnel(pfs, xlab = "OS")
dev.off()

#偏移检验
metabias(pfs,k.min=4,method="linreg")#k.min是最小文献纳入,method="linreg"是默认的Egger检验，目前主流推荐；Begg检验用“rank”