Bdtjtk

I have a large data frame where I have one column (Phylum) that has repeated names and 253 other columns (each with a unique name) that have counts of the Phylum column. I would like to sum the counts within each column that correspond to each Phylum.

This is a simplified version of what my data look like:

     Phylum    sample1    sample2    sample3 ...    sample253

1    P1        2          3          5              5

2    P1        2          2          10             2

3    P2        1          0          0              1

4    P3        10         12         3              1

5    P3        5          7          14             15

I have seen similar questions, but they are for fewer columns, where you can just list the names of the columns you want summed. I don't want to enter 253 unique column names.

I would like my results to look like this

    Phylum    sample1    sample2    sample3 ...    sample253

1   P1        4          5          15             7

2   P2        1          0          0              1

3   P3        15         19         17             16

I would appreciate any help. Sorry for the format of the question, this is my first time asking for help on stackoverflow (rather than sleuthing).

If your starting file looks like this (test.csv):

Phylum,sample1,sample2,sample3,sample253

P1,2,3,5,5

P1,2,2,10,2

P2,1,0,0,1

P3,10,12,3,1

P3,5,7,14,15

Then you can use group_by and summarise_each from dplyr:

read_csv('test.csv') %>% 

  group_by(Phylum) %>% 

  summarise_each(funs(sum))

(I first loaded tidyverse with library(tidyverse).)

Note that, if you were trying to do this for one column you can simply use summarise:

read_csv('test.csv') %>% 

  group_by(Phylum) %>% 

  summarise(sum(sample1))

summarise_each is required to run that function (in the above, funs(sum)) on each column.