Find combinations of genome without any package - r

Question

I want to find how many combinations of genome are found in a sequence. I mean for binary combinations: AA,AT,AG,AC,... 16 combinations like that;or for 3-elemented combinations ATG,ACG,... 64 combinations like that. I know how to do that with a package and I will write down it here. I want to create my own code to perform this
seqinr package is perfect on its job. That is the code that i used for;
install.packages('seqinr')
library(seqinr)
m = read.fasta(file='sequence.fasta')
mseq = m[[1]]
count(mseq,2) # gives how many binary combinations are found in the seq
count(mseq,3) # gives how many 3-elemented combinations are found in the seq

Answer 1

This is a slow way to do it. I am certain it is faster in the bioconductor package.
# some practice data
mseq = paste(sample(c("A", "C", "G", "T"), 1000, rep=T), collapse="")
# define a function called count
count = function(mseq, n){
# split the sequence into every possible sub sequence of length n
x = sapply(1:(nchar(mseq) - n + 1), function(i) substr(mseq, i, i+n-1))
# how many unique sub sequences of length R are there?
length(table(x))
}
Actually just checked and this is pretty much how they did it:
function (seq, wordsize, start = 0, by = 1, freq = FALSE, alphabet = s2c("acgt"),
frame = start)
{
if (!missing(frame))
start = frame
istarts <- seq(from = 1 + start, to = length(seq), by = by)
oligos <- seq[istarts]
oligos.levels <- levels(as.factor(words(wordsize, alphabet = alphabet)))
if (wordsize >= 2) {
for (i in 2:wordsize) {
oligos <- paste(oligos, seq[istarts + i - 1], sep = "")
}
}
counts <- table(factor(oligos, levels = oligos.levels))
if (freq == TRUE)
counts <- counts/sum(counts)
return(counts)
}
If you want to find the code for a function use getAnywhere()
getAnywhere(count)

Answer 2

The simple thing to do is just something like this:
# Generate a test sequence
set.seed(1234)
testSeq <- paste(sample(LETTERS[1:3], 100, replace = T), collapse = "")
# Split string into chunks of size 2 and then count occurrences
testBigram <- substring(testSeq, seq(1, nchar(testSeq), 2), seq(2, nchar(testSeq), 2))
table(testBigram)
AA AB AC BA BB BC CA CB CC
10 10 14 3 3 2 2 5 1
Here is a way using a "function factory" (https://adv-r.hadley.nz/function-factories.html).
The 2-element and 3-element combinations are n-grams of size 2 and 3. So we make this n-gram function factory.
# Generate a function to create a function
ngram <- function(size) {
function(myvector) {
substring(myvector, seq(1, nchar(myvector), size), seq(size, nchar(myvector), size))
}
}
# Assign the functions names (optional)
bigram <- ngram(2)
trigram <- ngram(3)
# 2 element combinations
table(bigram(testSeq))
AA AB AC BA BB BC CA CB CC
10 10 14 3 3 2 2 5 1
# count of 2 element combinations
length(unique(bigram(testSeq)))
[1] 9
# counting function
count <- function(mseq, n) length(unique(ngram(n)(mseq)))
count(testSeq, 2)
[1] 9
# and if we wanted to do with with 3 element combinations
table(trigram(testSeq))