Skip to content

matrix merging question

5 messages · arun, Elio Shijaku

arun
# 
HI Elio,
Not sure whether this helps:
length(list.files(pattern=".txt"))
#[1] 22
?lst1 <- lapply(list.files(pattern=".txt"),function(x) read.table(x,header=TRUE,stringsAsFactors=FALSE))
?sapply(lst1,dim)
?sapply(lst1,function(x) all(apply(rbind(colnames(x),rownames(x)),2,function(y) y[1]==y[2])))
?#[1] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
#[16] TRUE TRUE TRUE TRUE TRUE TRUE TRUE
lst2 <- lapply(lst1,as.matrix)
uNrownames <- unique(as.vector(sapply(lst2,rownames)))
res1 <- matrix(0,nrow=length(uNrownames), ncol=ncol(lst2[[1]]), dimnames=list(uNrownames,NULL))
?
for(i in seq_along(lst2)){
? mat1 <- lst2[[i]]
?res1[rownames(mat1),] <- res[rownames(mat1),] + mat1
?res1
?}

?dim(res1)
#[1] 90 90



A.K.
On Tuesday, February 25, 2014 2:52 PM, Elio Shijaku <selius at gmail.com> wrote:
Hi Arun,

I wanted to ask you again on the issue of merging several matrices. I tried to look through R-related pages on how to merge several symmetric matrices of same dimension (90x90) but with row names that need to be matched, but I didn't find any solution. What I found concerned the merging of two matrices which are not symmetric and that is not my case.

Can you help me with any code?

Attached are the files (in tab delimited text and/ Excel) that need to be merged. 

What I want is a 90x90 final matrix that is the sum (aggregate) of all 22 matrices and that matches each variable. 

I would appreciate your help if possible.

Best,

Elio
arun
# 
Hi,
Sorry, a typo:

?res1[rownames(mat1),] <- res[rownames(mat1),] + mat1
????????????????????????? ###should be:

res1[rownames(mat1),] <- res1[rownames(mat1),] + mat1

A.K.
On Tuesday, February 25, 2014 4:03 PM, arun <smartpink111 at yahoo.com> wrote:
HI Elio,
Not sure whether this helps:
length(list.files(pattern=".txt"))
#[1] 22
?lst1 <- lapply(list.files(pattern=".txt"),function(x) read.table(x,header=TRUE,stringsAsFactors=FALSE))
?sapply(lst1,dim)
?sapply(lst1,function(x) all(apply(rbind(colnames(x),rownames(x)),2,function(y) y[1]==y[2])))
?#[1] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
#[16] TRUE TRUE TRUE TRUE TRUE TRUE TRUE
lst2 <- lapply(lst1,as.matrix)
uNrownames <- unique(as.vector(sapply(lst2,rownames)))
res1 <- matrix(0,nrow=length(uNrownames), ncol=ncol(lst2[[1]]), dimnames=list(uNrownames,NULL))
?
for(i in seq_along(lst2)){
? mat1 <- lst2[[i]]
?res1[rownames(mat1),] <- res[rownames(mat1),] + mat1
?res1
?}

?dim(res1)
#[1] 90 90



A.K.
On Tuesday, February 25, 2014 2:52 PM, Elio Shijaku <selius at gmail.com> wrote:
Hi Arun,

I wanted to ask you again on the issue of merging several matrices. I tried to look through R-related pages on how to merge several symmetric matrices of same dimension (90x90) but with row names that need to be matched, but I didn't find any solution. What I found concerned the merging of two matrices which are not symmetric and that is not my case.

Can you help me with any code?

Attached are the files (in tab delimited text and/ Excel) that need to be merged. 

What I want is a 90x90 final matrix that is the sum (aggregate) of all 22 matrices and that matches each variable. 

I would appreciate your help if possible.

Best,

Elio
arun
# 
Hi,
I am not sure about your final result.? In your initial post, you mentioned "but with row names that need to be matched".? Anyway, this gives me symmetric matrix
##rownames are the same as colnames for each matrix, so:

uNrownames <- unique(as.vector(sapply(lst2,rownames)))
res <- matrix(0,nrow=length(uNrownames), ncol=length(uNrownames), dimnames=list(uNrownames,uNrownames))
?for(i in seq_along(lst2)){
?res[rownames(lst2[[i]]),rownames(lst2[[i]])] <- res[rownames(lst2[[i]]),rownames(lst2[[i]])] + lst2[[i]]
?res
?}


isSymmetric(res)
#[1] TRUE
dim(res)
#[1] 90 90


A.K.
On Tuesday, February 25, 2014 4:17 PM, Elio Shijaku <selius at gmail.com> wrote:
Hi Arun,

Yes, I figured that, thanks a lot for your help.

Howver, when I test res1 for symmetricity I get:
I need the final matrix to be symmetric.


Any idea?


E.
On Tue, Feb 25, 2014 at 10:14 PM, arun <smartpink111 at yahoo.com> wrote:
Hi,