Why heatmap.2 in R failed to read the numeric data frame?

Question

I have the following data

GOBPID  Term    col1 col2 col3 col4 col5
GO:0001659  temperature_homeostasis 3.49690559977475    0   0   0   0
GO:0001660  fever_generation    3.22606939096511    0   0   0   0

which I tried to read with heatmap.2 function:

library("gplots")
dat <- read.table("http://dpaste.com/1486837/plain/",header=TRUE)
dat   <- dat[,!names(dat) %in% c("GOBPID")];
dat2 <- dat[,-1]
rownames(dat2)<-dat[,1]
heatmap.2(dat2,symm=FALSE,trace="none");

In my understanding it should read the data.frame correctly. But why it failed?

Error in heatmap.2(dat, symm = FALSE, trace = "none") : 
  `x' must be a numeric matrix

Update: I find it strange because this work,

library("gplots")
round(Ca <- cor(attitude), 2)
heatmap.2(Ca, symm = FALSE, margins = c(6,6)) 

The structure of Ca is the same with my dat2, no?

Solution

Function heatmap.2() expects that x will be numeric matrix. So you need to convert your data frame dat2 to matrix with function as.matrix().

heatmap.2(as.matrix(dat2),symm=FALSE,trace="none")