如何知道一列中每个观测的频率并将它们按r排序？

Question

I have a column and each row is a string. I want to find 1.The frequency of each sequence 2.Sort the result by frequency from high to low 3.If the frequency are the same for multiple strings, I sort them by alphabet of the sequence.

My data looks like

   ID             seq
1   1 BBBBBBIRBBRBBBB
2   2 BBBBBBIRRRRRBBB
3   3 BBBBBBIRRRRRRRR
4   4 BBBBBBITBBBBBBB
5   5 BBBBBBITBBBRBBX
6   6 BBBBBBITTTTBBCX
7   7 BBBBBBITTTTTTTT
8   8 BBBBBBOBBBBBBTX
9   9 BBBBBBOBBBBBBXB
10 10 BBBBBBIRBBRBBBB
11 11 BBBBBBIRRRRRBBB
12 12 BBBBBBIRRRRRRRR
13 13 BBBBBBITBBBBBBB
14 14 BBBBBBITBBBRBBX
15 15 BBBBBBIRBBRBBBB
16 16 BBBBBBIRRRRRBBB
17 17 BBBBBBIRRRRRRRR
18 18 BBBBBBIRBBRBBBB
19 19 BBBBBBIRRRRRBBB
20 20 BBBBBBIRRRRRBBB

ID<-c(1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20)
seq<-c('BBBBBBIRBBRBBBB','BBBBBBIRBBRBBBB',  'BBBBBBIRRRRRBBB', 'BBBBBBIRRRRRRRR',  'BBBBBBITBBBBBBB',  'BBBBBBITBBBRBBX',  'BBBBBBITTTTBBCX',  'BBBBBBITTTTTTTT',  'BBBBBBOBBBBBBTX',  'BBBBBBOBBBBBBXB',  'BBBBBBIRBBRBBBB',  'BBBBBBIRRRRRBBB',  'BBBBBBIRRRRRRRR',  'BBBBBBITBBBBBBB',  'BBBBBBITBBBRBBX',  'BBBBBBIRBBRBBBB',  'BBBBBBIRRRRRBBB',  'BBBBBBIRRRRRRRR',  'BBBBBBIRBBRBBBB',  'BBBBBBIRRRRRBBB')
data.frame(ID,seq)

I want the result looks like this

sequence        Frequency
BBBBBBIRBBRBBBB 5
BBBBBBIRRRRRBBB 4
BBBBBBIRRRRRRRR 3
BBBBBBITBBBBBBB 2
BBBBBBITBBBRBBX 1
BBBBBBITTTTBBCX 1
BBBBBBITTTTTTTT 1
BBBBBBOBBBBBBTX 1
BBBBBBOBBBBBBXB 1

Thanks in advance!!

Answer 1

Can do this with data.table :

library(data.table)

setDT(df)[, .N, by = seq][order(-N)]

It is worth noting that data.table consistently beats dplyr in terms of speed on different sample sizes:

Number of the top is how many times original sample was repeated.

Here is the code to reproduce:

library(data.table)
library(dplyr)
dtWay <- function(ID, seq) {
  dt <- data.table(ID, seq);
  setkey(dt, seq);
  return(dt[, .N, by = seq][order(-N)])
}
dplyrWay <- function(ID, seq) {
  df <- data.frame(ID, seq)
  res <- df %>% 
    dplyr::group_by(seq) %>% 
    dplyr::summarize(frequency = length(ID)) %>% 
    dplyr::arrange(desc(frequency)) %>%
    dplyr::rename(sequence = seq)
  return (res)
}

N <- c(3, 4, 5, 6)
n <- 10^N

library(microbenchmark)
dev.off()
par( mfrow = c( 2, 2 ) )
res <- lapply(n, function(x) {

  ID <-c(1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19)
  ID <- rep(ID, times = x)
  seq<-c('BBBBBBIRBBRBBBB',  'BBBBBBIRRRRRBBB',   'BBBBBBIRRRRRRRR',  'BBBBBBITBBBBBBB',  'BBBBBBITBBBRBBX',  'BBBBBBITTTTBBCX',  'BBBBBBITTTTTTTT',  'BBBBBBOBBBBBBTX',  'BBBBBBOBBBBBBXB',  'BBBBBBIRBBRBBBB',  'BBBBBBIRRRRRBBB',  'BBBBBBIRRRRRRRR',  'BBBBBBITBBBBBBB',  'BBBBBBITBBBRBBX',  'BBBBBBIRBBRBBBB',  'BBBBBBIRRRRRBBB',  'BBBBBBIRRRRRRRR',  'BBBBBBIRBBRBBBB',  'BBBBBBIRRRRRBBB')
  seq  <- rep(seq, times = x)

  m <- microbenchmark( "data.table" = dtWay(ID, seq),
                       "dplyr" = dplyrWay(ID, seq),
                       times = 10, unit = "sc")

  a <- boxplot(m, main = x, xlab ="", ylab = "time")
})

Answer 2

If you want to exert more control over the sorting and names you could use the following dplyr functions.

library(dplyr)
# assumes df is a data frame with seq and ID columns
df %>% 
  group_by(sequence = seq) %>% 
  summarize(frequency = length(ID)) %>% 
  arrange(-frequency)

Answer 3

I like dplyr .

install.packages('dplyr')
library(dplyr)

df <- group_by(df, seq)
df <- count(df, seq)

count(df, seq)
Source: local data frame [9 x 2]

          seq     n
       (fctr) (int)
 1 BBBBBBIRBBRBBBB     4
 2 BBBBBBIRRRRRBBB     4
 3 BBBBBBIRRRRRRRR     3
 4 BBBBBBITBBBBBBB     2
 5 BBBBBBITBBBRBBX     2
 6 BBBBBBITTTTBBCX     1
 7 BBBBBBITTTTTTTT     1
 8 BBBBBBOBBBBBBTX     1
 9 BBBBBBOBBBBBBXB     1

That looks like your desired output, no? Not sure why it there's only 4 counts of the first sequence, though.