Dplyr select ends_with

Question

I'm starting with a df that looks like this:

  sample_id        target_id length     eff_length est_counts    tpm   class
1 SRR3884838C      (A)n     69          70          0        0.00000     0
2 SRR3884838C     (AC)n     69          70          0        0.00000     0
3 SRR3884838C     (AG)n     69          70          0        0.00000     0
4 SRR3884838C     (AT)n     69          70          5        15.98870    0

I would like to use dplyr's select function to select only sample_ids that end with the letter C, as well as the target_id and tpm.

Example data:

> dput(droplevels(head(te,4)))
structure(list(sample_id = structure(c(1L, 1L, 1L, 1L), .Label = "SRR3884838C", class = "factor"), 
target_id = structure(1:4, .Label = c("(A)n", "(AC)n", "(AG)n", 
"(AT)n"), class = "factor"), length = c(69L, 69L, 69L, 69L
), eff_length = c(70L, 70L, 70L, 70L), est_counts = c(0, 
0, 0, 5), tpm = c(0, 0, 0, 15.9887), class = c(0L, 0L, 0L, 
0L)), .Names = c("sample_id", "target_id", "length", "eff_length", 
"est_counts", "tpm", "class"), row.names = c(NA, 4L), class = "data.frame")

I tried using the following:

  teC <- select(te, (sample_id, ends_with("C")), target_id, tpm)

Which gives me the sample_id, target_id and tpm but does not select only the sample_id's that end with C eg:

      sample_id        target_id  tpm
9759  SRR3884843CxS   Tigger15a   0.00000e+00
9760  SRR3884843CxS   Tigger16a   0.00000e+00
9761  SRR3884843CxS   Tigger16b   0.00000e+00

Am I doing something wrong with select? I was able to work on example data from a tutorial site without issue.

Answer 1

select is used to keep variables (read: columns) by name, which is what you're doing with sample_id , target_id , and tmp . If you want to further filter by the values within sample_id , then add filter :

teC <- te %>% select(sample_id, target_id, tpm) %>% filter(grepl("C$", sample_id))

The regular expression "C$" will match strings that end with "C"; "CxS$" will match strings ending with "CxS"; and "(C|CxS)$" will match both.