combine unequal length matrices in loop with cbind()

Question

I have a simple question but it cost me hours. I would like to cbind() a matrix and a dataframe. The point is, they don't have equal lengths.

matrix:

condition
        [,1]
ILMN_1666845 TRUE

ILMN_1716400 TRUE

Data.frame

a

t1 t2 t3 t4 1 0 1 1 1

If I use cbind() without a loop, everything is ok and this is the result:

b<-cbind(condition,a) b
        condition t1  t2  t3  t4
ILMN_1666845 TRUE 0 1 1 1

ILMN_1716400 TRUE 0 1 1 1

But in a for loop I get the following error: Error in data.frame(..., check.names = FALSE) : arguments imply differing number of rows: 0, 1

Can anyone help me? Thanks!

For loop code:

for (p in 1:nrow(outcomes)) {

id <- apply(regulationtable, 1, function(i)

sum(i[1:length(regulationtable)] != outcomes[p,])==0)

idd<-as.matrix(id)

condition = subset(idd, idd[,1]==TRUE)

a<-as.data.frame(t(outcomes[p,]))

b<-cbind(condition,a)

write.table(b, "file.txt", append=TRUE)}

If you could provide an example with reproducible data, then it is much easier. The loop code for instance contain the variable outcomes which is unknown. — Christian Bøhlke
this is related to your previous question. Could you describe what exactly you're trying and what the structure is of both outcomes and regulationtable? If possible, provide us with a minimal example that reproduces the error. — Joris Meys
I created a minimal example that recreated the error you described. I also added a better way to avoid the trouble you had in this and your former question. If you need more explanation, just ask. — Joris Meys

Joris Meys Joris Meys · Accepted Answer · 2011-03-23T16:06:43

As far as I could read from your code, you try to cbind a possible empty object, which never works. That's also what the error is telling you. Probably at some point a is just empty, as there are no matches. So just add a condition

if(sum(id) !=0) { ... }

You could benefit quite a lot from rewriting your code to take this into account. I tried to guess what you wanted to do, and this code does exactly the same :

xx <- apply(outcomes,1,function(p){
    id <- apply(regulationtable,1,function(i)
      sum(i != p ) == 0)
    if(sum(id) !=0)
     cbind(as.data.frame(id[id]),t(p))
})

write.table(do.call(rbind,xx),file="file")

It returns you a list xx with, for every possible outcome, the genes that have the same regulationpattern. This is tested with :

outcomes <- expand.grid(c(0,1),c(0,1),c(0,1),c(0,1))

regulationtable <- data.frame(
    t1=sample(0:1,10,replace=T),
    t2=sample(0:1,10,replace=T),
    t3=sample(0:1,10,replace=T),
    t4=sample(0:1,10,replace=T)
)
rownames(regulationtable) <- paste("Gene",1:10,sep="-")

combine unequal length matrices in loop with cbind()

1 Answers