用每个样本列的特定值过滤掉行

Question

Keep rows (Obs) which Obs value is over the threshold value of the sample IN AT LEAST THREE SAMPLES. Remove rows which have 2 or less.

ie.

• Obs 1 has only S5 over the threshold so would be filteresd out;
• Obs2 has 4 and Obs 3 has 3 so they would remain in the df.

.

df <- data.frame(column=c("threshold", "Obs1", "Obs2", "Obs3"), S1 = c(1.7,1.4,1.9,1.3), S2= c(0.9,0.8,2,1), S3=c(2.5,2.4,2.1,0.5), S4=c(0.4,0.5,0.6,0.9), S5=c(1.2,1.4,1.3,1.6))
 df

    column      S1  S2  S3  S4  S5
    threshold  1.7 0.9 2.5 0.4 1.2 
    Obs1       1.4 0.8 2.4 0.5 1.4 
    Obs2       1.9 2.0 2.1 0.6 1.3
    Obs3       1.3 1.0 0.5 0.9 1.6

Desired output:

column      S1  S2  S3  S4  S5
 
Obs2       1.9 2.0 2.1 0.6 1.3
Obs3       1.3 1.0 0.5 0.9 1.6

I do not know how to code it but I wonder if using some logic like this:

logic <- if df (S1-5)>= threshold value then =1; if df (S1-5) < threhold then = 0 

library(dplyr)
logic  %>% rowwise %%
    filter(sum(c_across(where(is.numeric))) >= 3) %>%
    ungroup

Answer 1

If we are using rowwise with c_across , just slice the rows without the 'threshold' row and then do the comparison > with the corresponding slice d dataset with 'threshold' rows

library(dplyr)
df %>% 
    slice(-1) %>% 
    rowwise %>%
    filter(sum(c_across(where(is.numeric)) 
                > 
                  (df %>% 
                      slice(1) %>%
                   select(-1))) >=3) %>%
    ungroup

-output

# A tibble: 2 x 6
#  column    S1    S2    S3    S4    S5
#  <chr>  <dbl> <dbl> <dbl> <dbl> <dbl>
#1 Obs2     1.9     2   2.1   0.6   1.3
#2 Obs3     1.3     1   0.5   0.9   1.6

If there are other character columns as well, we can change the select for subset data

df  %>% 
   slice(-1) %>%
   rowwise %>% 
   filter(sum(c_across(where(is.numeric)) > df %>%
             slice(1) %>%
             select(where(is.numeric))) >=3)

---

Or another option with map

library(purrr)
library(magrittr)
i1 <- map(df %>%
              select(where(is.numeric)),  ~ .x[-1] >  first(.x)) %>% 
        reduce(`+`) %>% 
        is_greater_than(2)
df %>% 
     slice(-1) %>% 
     filter(i1)

---

Or using base R with rowSums

df[-1,][rowSums(df[-1, -1] > df[1, -1][col(df[-1, -1])]) >=3,]
#  column  S1 S2  S3  S4  S5
#3   Obs2 1.9  2 2.1 0.6 1.3
#4   Obs3 1.3  1 0.5 0.9 1.6

Answer 2

For future reference: If you work with columns which are character, you need to make sure the columns with the values are numeric, if not, transform them

df <- type.convert(df, as.is = TRUE) 

and then, this should work

df2 <- df %>% slice(-1) %>% rowwise %>% filter(sum(c_across(where(is.numeric)) > (df %>%slice(1))) >=3)