saving results from nest for loop - r

Question

My data looks something like this:
patient <- c(1,2,3,4,5)
outcome1 <- c(rnorm(5))
outcome2 <- c(rnorm(5))
outcome3 <- c(rnorm(5))
outcome4 <- c(rnorm(5))
outcome5 <- c(rnorm(5))
exposure1 <- c(rnorm(5))
exposure2 <- c(rnorm(5))
exposure3 <- c(rnorm(5))
exposure4 <- c(rnorm(5))
exposure5 <- c(rnorm(5))
covariate1 <- c(rnorm(5))
covariate2 <- c(rnorm(5))
data <- data.frame(patient <- patient,
outcome1 <- outcome1,
outcome2 <- outcome2,
outcome3 <- outcome3,
outcome4 <- outcome4,
outcome5 <- outcome5,
exposure1 <- exposure1,
exposure2 <- exposure2,
exposure3 <- exposure3,
exposure4 <- exposure4,
exposure5 <- exposure5,
covariate1 <- covariate1,
covariate2 <- covariate2)
I am using the following function to conduct a patrial correlation test and spit out the outcome. This function works great when subsetting a value at a time.
pcor.fit <- function(outcome, exposure, data, cov.columns){
temp <- pcor.test(data[,outcome], data[,exposure], as.matrix(data[,cov.columns]))
temp1 <- as.numeric(temp["estimate"])
temp2 <- as.numeric(temp["estimate"]/temp["statistic"]) ## se
temp3 <- as.numeric(temp["p.value"])
return(c(outcome = outcome, exposure = exposure, estimate=temp1, se=temp2, p=temp3))
}
The only problem is that I want to get partial combinations of all possible combinations of outcome a exposure. In this case it would be 25 (5 exposure and 5 outcomes). therefore I ran a loop to run through the combination of outcome and exposures, where outcome and exposures are lists of the variable names.
for (i in outcome) {
for (j in exposure) {
print(pcor.fit(outcome = i, exposure = j, data = data, cov.columns = covariates))
}
}
This works fine in printing the results, but how can I save the results of my function and loop? I assume I need to create an empty matrix first?

Answer 1

If I have understood correctly this answer would provide a reproducible question along with answer that you are looking for.
library(ppcor)
outcome <- grep('outcome', names(data), value = TRUE)
exposure <- grep('exposure', names(data), value = TRUE)
covariates <- grep('covariate', names(data), value = TRUE)
pcor.fit <- function(outcome, exposure, data, cov.columns){
temp <- pcor.test(data[,outcome], data[,exposure], as.matrix(data[,cov.columns]))
temp1 <- as.numeric(temp["estimate"])
temp2 <- as.numeric(temp["estimate"]/temp["statistic"]) ## se
temp3 <- as.numeric(temp["p.value"])
return(data.frame(outcome, exposure, estimate=temp1, se=temp2, p=temp3))
}
result <- vector('list', length(outcome) * length(exposure))
k <- 0
for (i in outcome) {
for (j in exposure) {
k <- k + 1
result[[k]] <- pcor.fit(outcome = i, exposure = j, data = data, cov.columns = covariates)
}
}
result <- do.call(rbind, result)
result
# outcome exposure estimate se p
#1 outcome1 exposure1 0.224018424 0.6891356 0.77598158
#2 outcome1 exposure2 0.615505519 0.5572939 0.38449448
#3 outcome1 exposure3 -0.555796882 0.5878307 0.44420312
#4 outcome1 exposure4 -0.261538517 0.6824945 0.73846148
#5 outcome1 exposure5 0.345310335 0.6636116 0.65468966
#6 outcome2 exposure1 -0.664104445 0.5286612 0.33589556
#7 outcome2 exposure2 -0.584807063 0.5735855 0.41519294
#...
#...
data
set.seed(123)
data <- data.frame(patient = c(1,2,3,4,5),
outcome1 = rnorm(5),
outcome2 = rnorm(5),
outcome3 = rnorm(5),
outcome4 = rnorm(5),
outcome5 = rnorm(5),
exposure1 = rnorm(5),
exposure2 = rnorm(5),
exposure3 = rnorm(5),
exposure4 = rnorm(5),
exposure5 = rnorm(5),
covariate1 = rnorm(5))