Correlation analysis with R "Hmisc" Package [spearman, pearson]

Hello, 

 

This is about correlation analysis among factors.

The correlation analysis used to lots of study area.

In my case, microbiome and enviornmental factors are analyzed by correlation methods.

 

 

 

"Hmisc" package is used in RStudio.

Let's follow the below steps to analysis.

 

 

 

 

 

 

1. Install packages

 

install.packages("Hmisc")

install.packages("xlsx")

 

 

 

 

 

 

 

2. Run packages

 

library(Hmisc)

library(xlsx)

 

 

 

 

 

 

 

 

3. Set calculate matrix

 

flattenCorrMatrix <- function(cormat, pmat) {

  ut <- upper.tri(cormat)

  data.frame(

    row = rownames(cormat)[row(cormat)[ut]],

    column = rownames(cormat)[col(cormat)[ut]],

    cor  =(cormat)[ut],

    p = pmat[ut]

  )

}

 

 

 

 

 

 

 

 

4. Input dataframe

 

test <- read.delim("clipboard", row.names=1) # Copy and paste your data from excel

 

test <- t(test)

 

 

 

 

 

 

 

 

 

 

5. Calculate Correlation

 

rcorr1<-rcorr(as.matrix(test), type=c("spearman"))

datacorS1 <- flattenCorrMatrix(rcorr1$r, rcorr1$P)

datacorS1$p_BH = p.adjust(datacorS1$p, method = "BH")

datacorS1$ABS = abs(datacorS1$cor)

datacorS2 <- subset(datacorS1, p_BH<0.0001)

datacorS3 <- subset(datacorS2, ABS>0.8)

Here you can change method with spearman to pearson.

These are different calculating equations each other.

Simply, spearman is rank-based on value.

So, highest value will be number 1 and following are 2, 3 ,4... etc.

In the other hands, raw value will be used in pearson method.

Finally, you can understand is it normalized (spearman) or not (pearson).

 

 

 

 

 

 

 

 

 

 

 

6. Make Output file

 

write.xlsx2(datacorS3, "E:/test.xlsx", sheetName = "Sheet1", col.name = T, row.names = F, append = F)

 

 

 

 

 

Final output will be same as above with excel.

cor is correlation value.

p is p-value. If you have lots of data, this correlation will be happen with multiple testing error.

So, q-value is necessary to adjust p-value.

p_BH is q-value. BH is Benjamin hochiberg who made modified p-value.

ABS is absolute value of correlation.

 

 

 

 

 

 

 

 

 

 

 

 

 

Supplementary story) 

Network visualization will be constructed with correlation analysis.

See the below.

These visualization will be posted soon.