如何将距离矩阵的下三角提取为R中的成对的列值

Question

I have a distance matrix called mydist . I want to extract the lower triangle of the matrix into pairwise combination of column values. For example:

sampleA          sampleB              values
S05-F13-P01_C   S05-F13-P01_C         2251
S08-F10-P01_C   S08-F10-P01_C         2246

. . so on

Data

mydist<-structure(c("2251", "1923", "2085", "1954", "2105", "0", "2246", 
"2094", "1955", "2127", "0", "0", "2521", "2110", "2329", "0", 
"0", "0", "2276", "2141", "0", "0", "0", "0", "2561"), .Dim = c(5L, 
5L), .Dimnames = list(c("S05-F13-P01_C", "S08-F10-P01_C", "S08-F11-P01_C", 
"S09-F66-P01_C", "S09-F67-P01_C"), c("S05-F13-P01_C", "S08-F10-P01_C", 
"S08-F11-P01_C", "S09-F66-P01_C", "S09-F67-P01_C")))

Answer 1

I would consider the following:

data.frame(as.table(mydist))[lower.tri(mydist, diag = TRUE), ]
##             Var1          Var2 Freq
## 1  S05-F13-P01_C S05-F13-P01_C 2251
## 2  S08-F10-P01_C S05-F13-P01_C 1923
## 3  S08-F11-P01_C S05-F13-P01_C 2085
## 4  S09-F66-P01_C S05-F13-P01_C 1954
## 5  S09-F67-P01_C S05-F13-P01_C 2105
## 7  S08-F10-P01_C S08-F10-P01_C 2246
## 8  S08-F11-P01_C S08-F10-P01_C 2094
## 9  S09-F66-P01_C S08-F10-P01_C 1955
## 10 S09-F67-P01_C S08-F10-P01_C 2127
## 13 S08-F11-P01_C S08-F11-P01_C 2521
## 14 S09-F66-P01_C S08-F11-P01_C 2110
## 15 S09-F67-P01_C S08-F11-P01_C 2329
## 19 S09-F66-P01_C S09-F66-P01_C 2276
## 20 S09-F67-P01_C S09-F66-P01_C 2141
## 25 S09-F67-P01_C S09-F67-P01_C 2561

Answer 2

We can try

i1 <- lower.tri(mydist, diag=TRUE)
i2 <- which(i1, arr.ind=TRUE)
data.frame(sampleA = colnames(mydist)[i2[,1]], 
        sampleB = colnames(mydist)[i2[,2]], value = mydist[i1])

Answer 3

This seems to work:

cbind(rownames(mydist)[which(lower.tri(mydist, diag=T), arr.ind=T)[,1]],
      colnames(mydist)[which(lower.tri(mydist, diag=T), arr.ind=T)[,2]],
      mydist[lower.tri(mydist, diag=T)])

Or, transforming it into a data.frame as @akrun does:

temp1 <-data.frame(sampleA=rownames(mydist)[which(lower.tri(mydist, diag=T), arr.ind=T)[,1]],
                   sampleB=colnames(mydist)[which(lower.tri(mydist, diag=T), arr.ind=T)[,2]],
                   value=as.numeric(mydist[lower.tri(mydist, diag=T)]), stringsAsFactors=F)