Print list in R without loops (apply) - r

Question

I hope that, this question will be nice tutorial for beginners in R (such as me).
I was used to programming languages where loops are necessary to manipulate the data, algorithms, etc.
Nevertheless, loops in R are slow, what can be seen in case of large data.
Fortunately R provides bulit-in functions which allow to iterating through elements and do some calculation in very efficient way.
Now I'd like to avoid loops when I'm analysying data in R. So I've read about lapply, apply and other useful functions.
I'd like to make correlation between first and each other column of my data and print: sample name, sample estimate and p-value in nice table - everything without for loop.
My idea - create fake data from stratch:
surv <- c(7.1,8,4,2,0.5,5,6)
geneA_expr <- runif(n = 7, min = 1, max = 30)
geneB_expr <- runif(n = 7, min = 1, max = 30)
geneC_expr <- runif(n = 7, min = 1, max = 30)
my_data <- data.frame(surv, geneA_expr, geneB_expr, geneC_expr)
Correlation test with apply - found it here in Stack Overflow and in manual:
md_stat <- apply(my_data[,2:4], 2, cor.test, my_data$surv, method="pearson")
md_stat is a list, now I'd like to visualize it nicely, but I have no idea how to do it, it's too complicated for me, so I used for loop
for(i in names(md_stat)){
cat(i ,md_stat[[i]]$estimate, md_stat[[i]]$p.value, '\n')
}
geneA_expr 0.2517658 0.5860052
geneB_expr 0.2438112 0.5982849
geneC_expr 0.8026801 0.02977544
How to replace above for loop by other bulit-in function?

Answer 1

unlist every list within md_stat. then bind the outputs into a matrix.
do.call(rbind, lapply(md_stat, unlist))

Answer 2

Try this
temp <- lapply(seq_along(md_stat), function(i) {
cat(names(md_stat)[[i]], md_stat[[i]]$estimate, md_stat[[i]]$p.value, '\n')
})

Answer 3

I can think of 4 ways for you to do this, 1 of which depends on the purrr package.
You could use a loop, walk from the purrr package, lapply and a recursive function.
library(microbenchmark)
library(purrr)
surv <- c(7.1,8,4,2,0.5,5,6)
geneA_expr <- runif(n = 7, min = 1, max = 30)
geneB_expr <- runif(n = 7, min = 1, max = 30)
geneC_expr <- runif(n = 7, min = 1, max = 30)
my_data <- data.frame(surv, geneA_expr, geneB_expr, geneC_expr)
md_stat <- apply(my_data[,2:4], 2, cor.test, my_data$surv, method="pearson")
md_loop <- function(md_stat) {
for(i in names(md_stat)){
cat(i ,md_stat[[i]]$estimate, md_stat[[i]]$p.value, '\n')
}
}
md_walk <- function(md_stat) {
walk(names(md_stat), function(i) {
cat(i ,md_stat[[i]]$estimate, md_stat[[i]]$p.value, '\n')
})
}
md_apply <- function(md_stat) {
lapply(names(md_stat), function(i) {
cat(i[[1]],md_stat[[i[[1]]]]$estimate, md_stat[[i[[1]]]]$p.value, '\n')
})
}
md_recursive <- function(md_stat) {
i <- names(md_stat)
if(length(i) < 1) {
NULL
} else {
cat(i[[1]],md_stat[[i[[1]]]]$estimate, md_stat[[i[[1]]]]$p.value, '\n')
md_recursive(tail(md_stat, -1))
}
}
md_speed <- microbenchmark(
md_loop(md_stat),
md_walk(md_stat),
md_apply(md_stat),
md_recursive(md_stat)
)
Speed comparisons