R/make_seqs_multinom_mod.R

In brouwern/compbio4all: Datasets and helpful functions for learning bioinformatics and computational biology

#' make_seqs_multinom_mod
#'
#' By Coghlan (2011)
#'
#' @param inputsequence inputsequence
#' @param X X
#'
#' @export

make_seqs_multinom_mod <- function(inputsequence, X)
{
  # Change the input sequence into a vector of letters
  #library(seqinr) # This function requires the SeqinR package.

  inputsequencevector <- seqinr::s2c(inputsequence)
  # Find the frequencies of the letters in the input sequence "inputsequencevector":
  mylength <- length(inputsequencevector)
  mytable <- table(inputsequencevector)
  # Find the names of the letters in the sequence
  letters <- rownames(mytable)
  numletters <- length(letters)
  probabilities <- numeric() # Make a vector to store the probabilities of letters
  for (i in 1:numletters)
  {
    letter <- letters[i]
    count <- mytable[[i]]
    probabilities[i] <- count/mylength
  }
  # Make X random sequences using the multinomial model with probabilities "probabilities"
  seqs <- numeric(X)
  for (j in 1:X)
  {
    seq <- sample(letters, mylength, replace=TRUE, prob=probabilities) # Sample with replacement
    seq <- seqinr::c2s(seq)
    seqs[j] <- seq
  }
  # Return the vector of random sequences
  return(seqs)
}