R-比较跨多列的组中的行

Question

I have a data frame containing samples (in rows) and their values in multiple columns. In some cases the sample has been repeated. What I want to do is compare the values in the columns for these repeats and put the output in a new df. If the values match I want to indicate this with a 1 and if they do not match a 0. NAs should result in NA.

What I try to do is similar to here . However, I only want to compare repeated samples, not all combinations of all rows as they do in the example in the link. But I cannot find a way to convert the solution given there to my problem.

Example data:

Sample  x.1  x.2  y.1  y.2  z.1  z.2
------------------------------------
ID1     66   66   102  104  33   37
ID2     66   72   100  104  31   35
ID2     66   72   100  104  NA   NA
ID3     64   66   104  104  35   37
ID4     72   72   100  102  31   37
ID4     72   72   NA   NA   31   37
ID4     72   72   100  102  31   31
ID5     66   66   102  102  35   35
ID5     66   72   100  100  31   37

Result I am looking for in a new df:

Sample  x.1  x.2  y.1  y.2  z.1  z.2
------------------------------------
ID2     1    1    1    1    NA   NA
ID4     1    1    NA   NA   1    0
ID5     1    0    0    0    0    0

I tried something along these lines but it did not work as it only give me 1 as an output, so that is definitely incorrect.

test <- df %>% 
  group_by(Sample) %>%
  mutate(across(1:6, funs(ifelse(.[1,]==.[2,], 1, 0))))

Answer 1

You can first remove groups which have only 1 row in each group and then summarise rest of the columns by group.

library(dplyr)

df %>%
  group_by(Sample) %>%
  filter(n() > 1) %>%
  summarise(across(x.1:z.2, 
            ~if(any(is.na(.))) NA else as.integer(n_distinct(.) == 1)))

#  Sample   x.1   x.2   y.1   y.2   z.1   z.2
#  <chr>  <int> <int> <int> <int> <int> <int>
#1 ID2        1     1     1     1    NA    NA
#2 ID4        1     1    NA    NA     1     0
#3 ID5        1     0     0     0     0     0