data.frame into matrix (R) , rownames() error

Question

I have a data.frame ("data.csv") of 93 x 28 I want to convert into matrix for further plotting, it looks like this:

SYMBOL  BT.20   CAL.51  MDA.MB.436  BT.549
A   3.039   4.908   3.865   3.818
B   4.349   5.399   6.071   5.313
C   7.509   8.091   6.48    6.660
D   3.429   4.394   3.622   3.873
E   3.369   6.716   3.557   3.346

the code

data <- read_csv("data.csv")
rnames <- data[,1]              # assign labels in column 1 to "rnames"  
mat_data <- data.matrix(data[,2:28])  # transform column 2 - end into a matrix
rownames(mat_data) <- rnames                  # assign row names

produces

ERROR: Error in `rownames<-`(`*tmp*`, value = list(SYMBOL = c("A", "B",  : 
  length of 'dimnames' [1] not equal to array extent

column 1 doesn't have duplicate names or missing values

edit: with read.csv() instead, it works just fine

Answer 1

The error is saying that the value being assigned to the rownames is a list and the "length" of a list with a single vector in it is 1, rather than the length of the vector. I'm not able to say why that should be happening since the usual behavior is for the "[" function to "drop" single columns into atomic vectors. Like you, I would have expected rnames to be an atomic vector of length 39, but it appears you need to use this instead:

rownames(mat_data) <- unlist( rnames )              # assign row names

Rich and Apom have better eyes than I do. The read_csv function is producing an object with a different class and associated extraction function than would have read.csv . Since you are in the "hadleyverse", you therefore should use this:

 rnames <- data[[1]]

The tibble -classed objects have a different version of the [ -function than do data.frames. The [[ -function appears to act the same for both classes.