CODEWITHSUNDEEP
rdata-wrangling
gdobbo
gdobbo

Reputation: 23

R grouping data by numeric numbers in a column

I am trying to group data by numbers in a column, I have tried different versions of group_by, cut, group etc but I have not been able to get it. I have a lot of data that looks like this:

  position variants

     3      snv
     5      snv
    12      snv
    17      mnv
    22 deletion
    27      snv
    33      snv
    35      snv
    42      snv
    46      mnv
    50      snv
    53 deletion
    60      snv
    62      snv
    65      snv
    70      snv

variants <- c(rep("snv", 3),rep("mnv", 1),rep("deletion", 1),rep("snv", 4), "mnv", rep("snv"), "deletion", rep("snv", 4))
variants              
position = c(3, 5, 12, 17, 22, 27, 33, 35, 42, 46, 50, 53, 60, 62, 65, 70)
position
patient1 = data.frame(position, variants)
patient1

I would like to be able to group the data something like this:

group  tally
1-10    2snv
11-20   1snv 1mnv
21-30   1deletion 1snv
31-40   2snv 
etc

so that i can run further downstream analysis. And be able to change it to groups of 1-5 or 1-2 etc. thank you very much

Upvotes: 2

Views: 1319

Answers (3)

akrun
akrun

Reputation: 887118

We can use tidvyerse to do a group by operation. Create a group of ranges with cut, summarise the frequency count based on the cut and the 'variants', then paste them together in summarise

library(dplyr)
patient1 %>% 
   group_by(group = cut(position, breaks = c(-Inf, seq(1, 100, 
       by = 10))), variants) %>%
   summarise(n = n()) %>%
   summarise(tally = paste(n, variants, collapse=' ', sep=""))

NOTE: Another option is findInterval which does similar option as cut but without the labels as it will output numeric index

Upvotes: 1

Ronak Shah
Ronak Shah

Reputation: 388982

In base R, you can create a group column using findInterval making groups of every 10 positions. We can then use aggregate and combine the count of variants with the variants to create one string for each group.

patient1$group <- with(patient1, findInterval(position, (seq(0, max(position), 10))))

aggregate(variants~group, patient1, function(x) {
  tb <- table(x)
  paste(tb, names(tb), collapse = ' ')
})

#  group         variants
#1     1            2 snv
#2     2      1 mnv 1 snv
#3     3 1 deletion 1 snv
#4     4            2 snv
#5     5      1 mnv 1 snv
#6     6 1 deletion 1 snv
#7     7            3 snv
#8     8            1 snv

Upvotes: 0

Ric
Ric

Reputation: 5722

Here a pure R solution. Of course you can replace variables by their corresponding calls:

variants <- c(rep("snv", 3),rep("mnv", 1),rep("deletion", 1),rep("snv", 4), "mnv", rep("snv"), "deletion", rep("snv", 4))
position = c(3, 5, 12, 17, 22, 27, 33, 35, 42, 46, 50, 53, 60, 62, 65, 70)
patient1 = data.frame(position, variants)

labels = cut(position, seq(0, max(position), 10))
groups = split(patient1 , labels)
lapply(groups , function(x) {
  paste( table(x$variants), names(table(x$variants)), collapse = ", " )
      }
  )

Upvotes: 2

Related Questions