如何使用dplyr对来自两个数据框的条件的列进行突变

Question

I have two dataframes, which are based on a third, larger dataset. I want to normalize the data in one dataframe according to the entries in the second dataframe - My favorite would be to use dplyr, but other packages/solutions are very appreciated, too :)

In my first dataframe, I have the counts of different organs.

Dataframe organ_count

 # A tibble: 5 x 2
                         organs count
                        <fctr> <int>
1                       Organ_A    23
2                       Organ_B    29
3                       Organ_C    24
4                       Organ_D    145
5                       Organ_E    97

In my second dataframe, I have the count of the same organs, but splitted upon in which state they appear in the large dataset I used as a source.

Dataframe organ_state_count

# A tibble: 15 x 3
              organs hmm_state count
             <fctr>     <chr> <int>
 1       Organ_A         E1     12
 2       Organ_A         E2     2
 3       Organ_A         E3     9
 4       Organ_B         E1     13
 5       Organ_B         E2     10
 6       Organ_B         E3     6
 7       Organ_C         E1     7
 8       Organ_C         E2     7
 9       Organ_C         E3     10
10       Organ_D         E1     72
11       Organ_D         E2     23
12       Organ_D         E3     50
13       Organ_E         E1     90
14       Organ_E         E2     2
15       Organ_E         E3     5

What I want to do now is:

I want to divide organ_state_count$count by the total number of entries for this organ (given in organ_state), resulting in the percentage of this organ for the given state.

I already tried something like this:

organ_state_count %>% 
    rowwise() %>% 
    do(organ_total = filter(organ_count,organs == .$organs)) %>%
    mutate(organ_norm=.$count/organ_total)

But it throws this error message:

Error in mutate_impl(.data, dots) : 
Evaluation error: arguments imply differing number of rows: 1, 0.
In addition: Warning messages:
1: Unknown or uninitialised column: 'count'. 
2: In Ops.factor(left, right) : ‘/’ not meaningful for factors

I must admit I'm fairly new to R and to the whole dplyr/tidyverse thing as well, so I'm a bit overwhelmed.

I also think that there is some kind of possibility of just using the organ_state_count frame for this task, and solve everything in just one dataframe, but I'm not sure how.

Thanks for your answers and help!

Answer 1

you can try something like:

df %>%
  group_by(organs) %>%
  mutate(tot = sum(count)) %>%
  ungroup() %>%
  mutate(pct = count/tot)

There's no need to use the first dataframe, as you have that information in the second dataframe already. Just select the columns you want to use for the final output.

data:

df <- read.table( text = "id organs hmm_state count
1 Organ_A E1 12
2 Organ_A E2 2
3 Organ_A E3 9
4 Organ_B E1 13
5 Organ_B E2 10
6 Organ_B E3 6
7 Organ_C E1 7
8 Organ_C E2 7
9 Organ_C E3 10
10 Organ_D E1 72
11 Organ_D E2 23
12 Organ_D E3 50
13 Organ_E E1 90
14 Organ_E E2 2
15 Organ_E E3 5", sep =" ", header = TRUE) 

output:

      id  organs hmm_state count   tot        pct
   <int>  <fctr>    <fctr> <int> <int>      <dbl>
1      1 Organ_A        E1    12    23 0.52173913
2      2 Organ_A        E2     2    23 0.08695652
3      3 Organ_A        E3     9    23 0.39130435
4      4 Organ_B        E1    13    29 0.44827586
5      5 Organ_B        E2    10    29 0.34482759
6      6 Organ_B        E3     6    29 0.20689655
7      7 Organ_C        E1     7    24 0.29166667
8      8 Organ_C        E2     7    24 0.29166667
9      9 Organ_C        E3    10    24 0.41666667
10    10 Organ_D        E1    72   145 0.49655172
11    11 Organ_D        E2    23   145 0.15862069
12    12 Organ_D        E3    50   145 0.34482759
13    13 Organ_E        E1    90    97 0.92783505
14    14 Organ_E        E2     2    97 0.02061856
15    15 Organ_E        E3     5    97 0.05154639