R计算两列数据的相关系数_数据相关性分析 correlation - R实现

a=log((rowMeans(tmp1))+1)
b=rowMeans(tmp2)
data=as.data.frame(cbind(a,b))
colnames(data)=c("valueA","valueB")
head(data)

library(ggplot2)
library(dplyr)
cor(a,b,method="pearson")
ggplot(data,aes(x=valueA,y=valueB))+ geom_point(size=1,shape=15)+geom_smooth(method=lm)

#install.packages("corrplot")
library(corrplot)
#matrix <- cor(data,method="spearman",use="everything")
matrix <- cor(data1,data2,method="spearman",use="complete.obs")
dim(matrix)
matrix[1:5,]
#correlation matrix
write.csv(matrix,file="data/spearman_correlation.csv",quote=F,row.names=T)

#correlation plot
png(file="corr_plot.png",width=3000,height=3000)
#数据少，半边的图
corrplot(matrix, type = "lower", order = "original", tl.col = "black", tl.srt = 45,cl.lim = c(0, 1))
#数据多，完整的图
#corrplot(matrix,title = "Correlation Plot",method = "color",outline = T, addgrid.col = "white", order="hclust", mar = c(0,0,1,0), addrect = 4, rect.col = "grey", rect.lwd = 1,cl.pos = "b", tl.col = "indianred4", tl.cex = 1, cl.cex = 3)
dev.off()

cor_JC<-as.numeric(matrix)
#如果是组内数据两两比较，会出现很多自己和自己比然后完全相关，结果=1的数值，记得移除（但不要移走其他1）
#cor_JC[1:5]
#rm=which(cor_JC==1)  #remove cor 1
#cor_JC=cor_JC[-rm]
#cor_JC[1:5]
cor_JC[cor_JC >0.95]
sum(cor_JC >0.95)
cor_JC[cor_JC < -0.95]
sum(cor_JC < -0.95)

#INPUT DATA
metagenomics[1:2,]

#COORELATION
cor <- 0
d<-c("0","0","0")
end=dim(metagenomics)[2]-1
for (i in 1:end) {
    x = metagenomics[,i]
    s=i+1
    for (j in s:dim(metagenomics)[2]) {
        cor_n <- cor(x, y = metagenomics[,j], method = "pearson")
        cor<-append(cor,cor_n)
        if(as.numeric(as.character(cor_n))>0.9){
            metab1<-colnames(metagenomics)[i]
            metab2<-colnames(metagenomics)[j]
            re<-c(metab1,metab2,cor_n)
            d<-rbind(d,re)
        }
    }
}

d<-as.data.frame(d)
colnames(d)<-c("metab1","metab2","pearson_cor")
d<-d[-1,]
d[1:5,]

cor<-cor[-1]
summary(cor)
length(cor[cor >0.8])
length(cor[cor >0.9])
#frequency
hist(cor,breaks=100,xlim=c(-1,1))
abline(v = 0.9, col = "red")

data<-as.data.frame(read.table("data/metabolomics_metagenomics_34JC.csv",sep=",",header=T,row.names=1,stringsAsFactors=F))
data<-t(data[-c(1,2),])
metagenomics<-data[,1:169]
metabolites<-data[,170:295]

#shuffles label in one group
j = sample(1:nrow(metabolites),1*nrow(metabolites)) 
#metabolites = metabolites[j,] 
#metagenomics = metagenomics[j,]

#metagenomics[1:2,]
#metabolites[1:2,]

cor <- numeric(dim(metagenomics)[2])
for (i in 1:dim(metagenomics)[2]) {
cor[i] <- cor(x = metabolites[,1], y = metagenomics[,i], use = "everything", method = "spearman")
}
#med
hist(cor,breaks=100)
#inter <- quantile(med, c(0.025,0.975))
#inter