Combine similar rows across two data frames - r

Question

Still getting the gist of R. I have two data frames where the rows are named with different coordinates (e.g. x_1013y_41403; see below). The coordinates form sets of five, each set makes a cross if plotted onto a grid. The center coordinate is in one data frame, and the four peripheral coordinates are in the other.
Center A B C D E F
x_723y_6363.txt 554 NA 604 NA 645 NA
x_749y_41403.txt 14 NA 6 NA 13 NA
Peripheral A B C D E F
x_1013y_41403.txt NA 1 NA 0 NA 0
x_459y_6363.txt NA 2 NA 1 NA 4
x_485y_41403.txt NA 0 NA 0 NA 0
x_723y_6100.txt NA 1 NA 0 NA 3
x_723y_6627.txt NA 1 NA 0 NA 1
x_749y_41139.txt NA 1 NA 0 NA 0
x_749y_41667.txt NA 2 NA 0 NA 0
x_987y_6363.txt NA 1 NA 0 NA 0
To form a set, the peripheral coordinates would have the same x or y location as the center coordinate. For example, the center coordinate x_723y_6363 would be associated with x_723y_6100 and x_723y_6627 (same x location), as well as x_459y_6363 and x_987y_6363 (same y location).
I would like to combine the coordinates into their respective sets, and name the set with the center coordinate. For the case above, I would end up with two rows, where each row is the summation of a set.
A B C D E F
x_723y_6363.txt 554 5 604 1 645 8
x_749y_41403.txt 14 4 6 0 13 0
I am not sure at all how this can be done. I have thought about creating regular expressions to pick out the x and y coordinates individually and then doing a comparison across the two data frames. Any help would be greatly appreciated!

Answer 1

I hope someone else comes up with a better answer as this is ugly. I would first split the .txt names into x and y values then loop over each of the variables that is NA in center and sum all values that are share an x or y value with that center. Edit: Changed the sapply to make it slightly nicer.
center <- read.table(textConnection("
A B C D E F
x_723y_6363.txt 554 NA 604 NA 645 NA
x_749y_41403.txt 14 NA 6 NA 13 NA"),
header = TRUE)
peripheral <- read.table(textConnection("
A B C D E F
x_1013y_41403.txt NA 1 NA 0 NA 0
x_459y_6363.txt NA 2 NA 1 NA 4
x_485y_41403.txt NA 0 NA 0 NA 0
x_723y_6100.txt NA 1 NA 0 NA 3
x_723y_6627.txt NA 1 NA 0 NA 1
x_749y_41139.txt NA 1 NA 0 NA 0
x_749y_41667.txt NA 2 NA 0 NA 0
x_987y_6363.txt NA 1 NA 0 NA 0"),
header = TRUE)
xpat <- "^([^y]+).*"
ypat <- ".*(y_[0-9]+)\\.txt"
center$x <- gsub(xpat, "\\1", rownames(center))
center$y <- gsub(ypat, "\\1", rownames(center))
peripheral$x <- gsub(xpat, "\\1", rownames(peripheral))
peripheral$y <- gsub(ypat, "\\1", rownames(peripheral))
vars <- c("B", "D", "F")
center[vars] <- sapply(peripheral[vars], function(col)
apply(center, 1, function(row) sum(col[peripheral$x %in% row["x"] | peripheral$y %in% row["y"]]) )
)
R> center
A B C D E F x y
x_723y_6363.txt 554 5 604 1 645 8 x_723 y_6363
x_749y_41403.txt 14 4 6 0 13 0 x_749 y_41403

Answer 2

Another option:
# function to split coordinates x and y:
f <- function(DF) structure(
t(sapply(strsplit(row.names(DF), "[_y.]"), `[`, c(2,4))),
dimnames=list(NULL, c("x", "y")))
# get x and y for peripheral data:
P <- cbind(Peripheral, f(Peripheral))
# get x and y for centers, and mark ids:
C <- cbind(Center, f(Center), id=1:nrow(Center))
# matching:
Q <- merge(merge(P, C[,c("x","id")], all=TRUE), C[,c("y","id")], by="y", all=TRUE)
# prepare for union:
R <- within(Q, {id <- ifelse(is.na(id.y), id.x, id.y); id.x <- NULL; id.y <- NULL})
# join everything and aggregate:
S <- rbind(R, C)
aggregate(S[,3:8], by=list(id=S$id), FUN=sum, na.rm=TRUE)
Result:
id A B C D E F
1 1 554 5 604 1 645 8
2 2 14 4 6 0 13 0