CODEWITHSUNDEEP
rcorrelation
user2300940
user2300940

Reputation: 2395

get output of cor.test as data frame

I am doing a correlation test for all colnames in matrix.dge.cpm.t against a vector samplesheet$cond1. I would like to extract the results of the test for all columns in matrix.dge.cpm.t which are tested.

cor.test <- apply(matrix.dge.cpm.t,2, function(col)cor.test(col,log(samplesheet$cond1,2)))

> matrix.dge.cpm.t[,1:5][1:5,]
  LOC117736868     naf1    vegfc    tenm3     dctd
1     6.519403 4.930011 4.223574 7.088435 5.065795
2     6.504621 5.113774 4.139711 7.149168 4.702029
3     6.424178 4.903654 4.007805 7.111132 4.822668
4     6.585717 4.861847 3.716316 7.170937 5.025178
5     6.442080 5.048240 4.078584 7.110264 4.635633

Output:

> head(cor.test )
$LOC117736868

    Pearson's product-moment correlation

data:  col and log(samplesheet$cond1, 2)
t = 1.7197, df = 30, p-value = 0.09579
alternative hypothesis: true correlation is not equal to 0
95 percent confidence interval:
 -0.05486633  0.58694383
sample estimates:
      cor 
0.2995587


$naf1

    Pearson's product-moment correlation

data:  col and log(ss.Likith.numeric.order$Cortisol, 2)
t = 0.78998, df = 30, p-value = 0.4357
alternative hypothesis: true correlation is not equal to 0
95 percent confidence interval:
 -0.2167293  0.4681440
sample estimates:
      cor 
0.1427528

Summary looks like this:

> head(summary(cor.test))
             Length Class Mode
LOC117736868 9      htest list
naf1         9      htest list
vegfc        9      htest list
tenm3        9      htest list
dctd         9      htest list
cep44        9      htest list

Examples of output I want:

LOC117736868 p-value = 0.09579 cor = 0.2995587

Upvotes: 1

Views: 1359

Answers (1)

obruzzi
obruzzi

Reputation: 556

Try this

# Create dataframe where to save results
res <- data.frame(matrix(nrow = 0, ncol = 4))
colnames(res) <- c("var1", "var2", "correlation", "pvalue")

# Correlation in loop
for(i in colnames(matrix.dge.cpm.t[,3:5])) {
  for(j in colnames(matrix.dge.cpm.t[,3:5])) {
    a <- cor.test(matrix.dge.cpm.t[[i]], matrix.dge.cpm.t[[j]])
    res <- rbind(res, data.frame(
      "var1" = i,
      "var2" = j,
      "correlation" = a$estimate,
      "pvalue" = a$p.value) )
  }
}

# Remove rownames
rownames(res) <- NULL

Another option is to use the Hmisc package and code presented here.

# ++++++++++++++++++++++++++++
# flattenCorrMatrix
# ++++++++++++++++++++++++++++
# cormat : matrix of the correlation coefficients
# pmat : matrix of the correlation p-values
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]
    )
}

library(Hmisc)
res <- rcorr(as.matrix(matrix.dge.cpm.t[,1:5]))
flattenCorrMatrix(res$r, res$P)

EDIT 8th October 2020

To run the matrix against one variable

# Create dataframe where to save results
res <- data.frame(matrix(nrow = 0, ncol = 3))
colnames(res) <- c("var1", "correlation", "pvalue")

# Correlation in loop
for(i in colnames(matrix.dge.cpm.t[,3:5])) {
    a <- cor.test(matrix.dge.cpm.t[[i]], samplesheet$cond1)
    res <- rbind(res, data.frame(
      "var1" = i,
      "correlation" = a$estimate,
      "pvalue" = a$p.value) )
}

Upvotes: 3

Related Questions