I am new to R and I have a question about extracting data from a list with multiple groups. For example, I have a set of data like this:
data(iris)
iris$Group = rep(c("High","Low", each=5))
iris = iris[sample(nrow(iris)),]
mylist = list(iris[1:50,], iris[51:100,], iris[101:150,])
head(mylist)[[1]]
Sepal.Length Sepal.Width Petal.Length Petal.Width Species Group
51 7.0 3.2 4.7 1.4 versicolor High
123 7.7 2.8 6.7 2.0 virginica High
147 6.3 2.5 5.0 1.9 virginica Low
23 4.6 3.6 1.0 0.2 setosa High
120 6.0 2.2 5.0 1.5 virginica Low
141 6.7 3.1 5.6 2.4 virginica High
Within each list, I would like to group by Species and calculate the P value by t.test of Sepal.Length between Group High and Low. For example, I would like to get the P value of between Group High and Low of Species virginica, and so on for each list.
I am confused about this. Could anyone help? Thanks!
In base R you can do the following
lapply(mylist, function(x)
with(x, t.test(Sepal.Length[Group == "High"], Sepal.Length[Group == "Low"])$p.value))
#[[1]]
#[1] 0.2751545
#
#[[2]]
#[1] 0.5480918
#
#[[3]]
#[1] 0.864256
Or a purrr/tidyverse approach
library(tidyverse)
bind_rows(mylist, .id = "id") %>%
group_by(id) %>%
nest() %>%
mutate(pval = map_dbl(data, ~t.test(
.x$Sepal.Length[.x$Group == "High"],
.x$Sepal.Length[.x$Group == "Low"])$p.value))
## A tibble: 3 x 3
# id data pval
# <chr> <list> <dbl>
#1 1 <tibble [50 × 6]> 0.275
#2 2 <tibble [50 × 6]> 0.548
#3 3 <tibble [50 × 6]> 0.864
To perform t-tests of Sepal.Length between Group = "Low" and Group = "High" within Species you can do
lapply(mylist, function(x)
with(x, setNames(sapply(unique(Species), function(y)
t.test(Sepal.Length[Group == "High" & Species == y], Sepal.Length[Group == "Low" & Species == y])$p.value), unique(Species))))
#[[1]]
#versicolor virginica setosa
#0.80669755 0.07765262 0.47224383
#
#[[2]]
# setosa virginica versicolor
# 0.6620094 0.2859713 0.2427945
#
#[[3]]
#versicolor setosa virginica
# 0.5326379 0.6412661 0.5477179
Keep in mind that you will have to adjust raw p-values for multiple hypothesis testing.
To account for multiple hypothesis testing, you could modify above code slightly to give
lapply(mylist, function(x)
with(x, p.adjust(setNames(sapply(unique(Species), function(y)
t.test(Sepal.Length[Group == "High" & Species == y], Sepal.Length[Group == "Low" & Species == y])$p.value), unique(Species)))))
#[[1]]
#versicolor virginica setosa
# 0.9444877 0.2329579 0.9444877
#
#[[2]]
# setosa virginica versicolor
# 0.7283836 0.7283836 0.7283836
#
#[[3]]
#versicolor setosa virginica
# 1 1 1
Here we use p.adjust with the default Holm correction.
Collected from the Internet
Please contact [email protected] to delete if infringement.
Comments