如何根据为行子集指定的阈值对列进行子集化

Question

I want to subset columns of a big dataframe that adhere to this rule:

For each row ( except row A) the value should be below 5.

Given the following example dataframe, I want the function to return a dataframe with only column c1 since all values in row B:E are below 5. I think the select_if function is probably the way to go, but I can't figure out how to exclude specific rows in that function.

Gene <- c("A", "B", "C", "D", "E")
c1 <- c(500, 1, 0, 3, 0)
c2 <- c(240, 235, 270, 100, 1)
c3 <- c(0, 3, 1000, 900, 2)
df1 <- data.frame(Gene, c1, c2, c3)
head(df1)

  Gene  c1  c2   c3
1    A 500 240    0
2    B   1 235    3
3    C   0 270 1000
4    D   3 100  900
5    E   0   1    2

Answer 1

A tidyverse solution is

df1 %>% 
  select(
    df1 %>% 
      filter(row_number() > 1) %>% 
      summarise(across(starts_with("c"), max)) %>% 
      pivot_longer(everything()) %>% 
      filter(value < 5) %>% 
      pull(name)
  )
   c1
1 500
2   1
3   0
4   3
5   0

Explanation: the code inside the select calculates the maximum value for each column after ignoring the first row. The result is then pivot ed into long format, creating default columns name and value . This data frame is filtered to select only those columns where every value is less than five. The name column is then pull ed and used as an argument to the outer select .

If you need other columns, just modify the select , for example,

df1 %>% 
  select(
    c("Gene", 
    df1 %>% 
      filter(row_number() > 1) %>% 
      summarise(across(starts_with("c"), max)) %>% 
      pivot_longer(everything()) %>% 
      filter(value < 5) %>% 
      pull(name)
  )
)

Answer 2

A base R solution is simple to code.

i <- sapply(df1[-1], \(x) all(x[-1] < 5))
df1[c(TRUE, i)]
#>   Gene  c1
#> 1    A 500
#> 2    B   1
#> 3    C   0
#> 4    D   3
#> 5    E   0

^{Created on 2022-06-03 by the reprex package (v2.0.1)}

Answer 3

To avoid reshaping or looping, use the colSums .

df1[c(1, which(colSums(df1[-1, -1] < 5) == 4) + 1)]
#   Gene  c1
# 1    A 500
# 2    B   1
# 3    C   0
# 4    D   3
# 5    E   0