CODEWITHSUNDEEP
rdplyr
neversaint
neversaint

Reputation: 64064

How to calculate t.test on grouped data based on given list of vectors with two values

I have the following data frame:


library(tidyverse)

dat <- tribble(
 ~sample, ~rep, ~area,
 "S1", "01", 100,
 "S1", "02", 200,
 "S1", "01", 300,
 "S2", "01", 10,
 "S2", "02", 20,
 "S2", "02", 30,
 "S3", "01", 1.2,
 "S3", "02", 2.3,
 "S3", "02", 3.1
)

dat
#> # A tibble: 9 x 3
#>   sample   rep  area
#>    <chr> <chr> <dbl>
#> 1     S1    01 100.0
#> 2     S1    02 200.0
#> 3     S1    01 300.0
#> 4     S2    01  10.0
#> 5     S2    02  20.0
#> 6     S2    02  30.0
#> 7     S3    01   1.2
#> 8     S3    02   2.3
#> 9     S3    02   3.1

What I want to do is to create a function that takes a list of vectors with a paired value and then calculate the P-value derived from t.test of area column.

Fore example given this list:

wanted_pairs <- list(c("S1","S2"), c("S2","S3"))

We would like to generate this data frame:

  S1 S2  0.0878746
  S2 S3  0.07564237

Manually, the above P-values are calculated with:

t.test(dat[dat$sample=="S1",]$area, dat[dat$sample=="S2",]$area)$p.value
t.test(dat[dat$sample=="S2",]$area, dat[dat$sample=="S3",]$area)$p.value

How can I achieve that?

Upvotes: 1

Views: 56

Answers (1)

akrun
akrun

Reputation: 887831

We can use a loop

library(dplyr)
lapply(wanted_pairs, function(x) 
      dat %>% 
         filter(sample %in% x) %>% 
         summarise(sample1 = unique(sample)[1], sample2 = unique(sample)[2],
                    pval = t.test(area[sample == sample1], 
                              area[sample == sample2])$p.value)) %>%
      bind_rows()
# A tibble: 2 x 3
#  sample1 sample2      pval
#    <chr>   <chr>     <dbl>
#1      S1      S2 0.0878746
#2      S2      S3 0.0897509

Upvotes: 1

Related Questions