[R] extract columns of a matrix/data frame

[Marc Schwartz]

On Tue, 2007-07-31 at 10:35 -0700, yuvika wrote:
> Hello all,
>    
>   I have a matrix whose column names look like
>    
>   a1  a2  b1  b2  b3  c1 c2
>   1   2    3    7    1    3   2
>   4   6    7    8    1    4   3
>    
>   Now, I can have any number of a's. not just two as shown above and
> same goes for b's and c's.  I need to extract all the a's columns and
> put them in another matrix, extract all b's columns and put them in
> some matrix and same goes for "c". How can I identify such pattern and
> get subsets of this matrix depending on columns names?
>    
>   I will appreciate a quick reply.
>   Thanks  a lot.


If 'MAT' is your matrix:

> MAT
     a1 a2 b1 b2 b3 c1 c2
[1,]  1  2  3  7  1  3  2
[2,]  4  6  7  8  1  4  3


You can use:

> sapply(letters[1:3], function(x) MAT[, grep(x, colnames(MAT))])
$a
     a1 a2
[1,]  1  2
[2,]  4  6

$b
     b1 b2 b3
[1,]  3  7  1
[2,]  7  8  1

$c
     c1 c2
[1,]  3  2
[2,]  4  3


which returns a list containing the three matrices as a consequence of
subsetting 'MAT" based upon the colnames.

This uses sapply() to loop over letters[1:3], which is:

> letters[1:3]
[1] "a" "b" "c"

and then uses grep() to get the indices of the colnames matching the
individual letters, passed as 'x' in each iteration of the sapply()
loop. For example:

> grep("a", colnames(MAT))
[1] 1 2


You can then manipulate each sub-matrix in the list as you require.

See ?sapply, ?grep and ?letters

HTH,

Marc Schwartz