CODEWITHSUNDEEP
rdataframematchcol
Letin
Letin

Reputation: 1275

Conditions to match all columns with one column

I have a data frame (df) where in I want to match each column with the last column in order to provide new values to each of those columns.

Here is my example data frame (df):

> df
              S1  S2  S3  S4  S5  main
Gene1         1   1   1   1   2   1
Gene2         1   2   1   1   1   1
Gene3         1   1   1   1   2   2
Gene4         2   1   1   1   1   1
Gene5         1   2   1   2   1   1
Gene6         1   1   1   1   1   2
Gene7        NA  NA   2   1   1   1
Gene8         1   2   1   1   1   2
Gene9         2   1   1   2   1   1

I want to match each of my columns from 1 to 5 with the last column with the following conditions. 'S' below refers to each column from 1 to 5.

If S = 2 and main = 2, then value is True Positive (TP)
If S = 2 and main = 1, then value is False Positive (FP)
If S = 1 and main = 2, then value is False Negative (FN)
If S = 1 and main = 1, then value is True Negative (TN)
And NAs to remain as NAs.

And therefore my new data frame (df_updated) should look like this.

> df_updated
              S1  S2  S3  S4  S5
Gene1         TN  TN  TN  TN  FP
Gene2         TN  FP  TN  TN  TN
Gene3         FN  FN  FN  FN  TP
Gene4         FP  TN  TN  TN  TN
Gene5         TN  FP  TN  FP  TN
Gene6         FN  FN  FN  FN  FN
Gene7         NA  NA  FP  TN  TN
Gene8         FN  TP  FN  FN  FN
Gene9         FP  TN  TN  FP  TN

I am aware of the match functions, but I am not sure how to loop them and use these above specific matches for each of the columns.

Any help appreciated, Thank you.

Upvotes: 2

Views: 172

Answers (2)

Ronak Shah
Ronak Shah

Reputation: 388982

Using base R, you can also create a function with nested ifelse and apply the function to every column and get values.

get_value <- function(x,main) {
 ifelse(main == 2 & x == 2, "TP", 
      ifelse(main == 1 & x == 2, "FP", 
            ifelse(main == 2 & x == 1, "FN", 
                 ifelse(main == 1 & x == 1 ,"TN", NA))))
}

df1 <- df[-ncol(df)]
df1[] <- lapply(df1, get_value, df$main)   

df1
#        S1   S2 S3 S4 S5
#Gene1   TN   TN TN TN FP
#Gene2   TN   FP TN TN TN
#Gene3   FN   FN FN FN TP
#Gene4   FP   TN TN TN TN
#Gene5   TN   FP TN FP TN
#Gene6   FN   FN FN FN FN
#Gene7 <NA> <NA> FP TN TN
#Gene8   FN   TP FN FN FN
#Gene9   FP   TN TN FP TN

Upvotes: 2

Joris C.
Joris C.

Reputation: 6234

You could use dplyr's case_when:

library(dplyr)

mutate_all(df, ~case_when(
           .x < main ~ "FN",
           .x > main ~ "FP",
           near(.x, 1) & near(.x, main) ~ "TN",
           near(.x, 2) & near(.x, main) ~ "TP"
           )) %>%
select(-main)
#>     S1   S2 S3 S4 S5
#> 1   TN   TN TN TN FP
#> 2   TN   FP TN TN TN
#> 3   FN   FN FN FN TP
#> 4   FP   TN TN TN TN
#> 5   TN   FP TN FP TN
#> 6   FN   FN FN FN FN
#> 7 <NA> <NA> FP TN TN
#> 8   FN   TP FN FN FN
#> 9   FP   TN TN FP TN

Upvotes: 3

Related Questions