Question

Input file:

df1 <- data.frame(row.names=c("w","x","y","z"), 
                  A=c(0,0,0,0),
                  B=c(0,1,0,0), 
                  C=c(1,0,1,0), 
                  D=c(1,1,1,1))

  A B C D
w 0 0 1 1
x 0 1 0 1
y 0 0 1 1
z 0 0 0 1

I want to apply an equation i.e. multiply row w to row x to get the pairwise value for w-x pair, as follows:

      A B C D
    w 0 0 1 1
X   x 0 1 0 1
--------------
   wx 0 0 0 1

to get row-wise analysis for w-x, w-y, w-y, w-z, x-y, x-z, y-z. and generate a new dataframe with 6 columns (two row names followed by the multiplied values).

That’s

w x 0 0 0 1
w y 0 0 1 1
w z 0 0 0 1
x y 0 0 0 1
x z 0 0 0 1
y z 0 0 0 1

Thanks.

Answer 1

If you don’t want the combo names in the resulting object, then we can combine elements of @DWin’s and @Owen’s Answers to provide a truly vectorised approach to the problem. (You can add the combination names as row names with one extra step at the end.)

First, the data:

dat <- read.table(con <- textConnection("  A B C D
w 0 0 1 1
x 0 1 0 1
y 0 0 1 1
z 0 0 0 1
"), header=TRUE)
close(con)

Take the combn() idea from @DWin’s Answer but use it on the row indices of dat:

combs <- combn(seq_len(nrow(dat)), 2)

The rows of combs now index the rows of dat that we want to multiply together:

> combs
     [,1] [,2] [,3] [,4] [,5] [,6]
[1,]    1    1    1    2    2    3
[2,]    2    3    4    3    4    4

Now we take the idea @Owen showed, namely dat[i, ] * dat[j, ] with i and j being the first and second rows of combs respectively. We convert to a matrix with data.matrix() as this will be more efficient for large objects, but the code will work with dat as a data frame too.

mat <- data.matrix(dat)
mat[combs[1,], ] * mat[combs[2,], ]

which produces:

> mat[combs[1,], ] * mat[combs[2,], ]
  A B C D
w 0 0 0 1
w 0 0 1 1
w 0 0 0 1
x 0 0 0 1
x 0 0 0 1
y 0 0 0 1

To see how this works, note that mat[combs[k,], ] produces a matrix with various rows repeated in the order specified by the combinations:

> mat[combs[1,], ]
  A B C D
w 0 0 1 1
w 0 0 1 1
w 0 0 1 1
x 0 1 0 1
x 0 1 0 1
y 0 0 1 1
> mat[combs[2,], ]
  A B C D
x 0 1 0 1
y 0 0 1 1
z 0 0 0 1
y 0 0 1 1
z 0 0 0 1
z 0 0 0 1

To get exactly what the OP posted, we can modify the rownames using a second combn() call:

> out <- mat[combs[1,], ] * mat[combs[2,], ]
> rownames(out) <- apply(combn(rownames(dat), 2), 2, paste, collapse = "")
> out
   A B C D
wx 0 0 0 1
wy 0 0 1 1
wz 0 0 0 1
xy 0 0 0 1
xz 0 0 0 1
yz 0 0 0 1