Nothing
#' Check and filter invalid base of nucleotide sequences
#'
#' In general, whitespaces and non-base characters are removed and characters are converted to uppercase in given method.
#'
#' @param ntseq Sequence (5' to 3') of one strand of the DNA nucleic acid duplex
#' as string or vector of characters
#'
#' @param method
#'
#' TM_Wallace: check and return "A","B","C","D","G","H","I","K","M","N","R","S","T","V","W" and "Y"
#'
#' TM_GC: check and return "A","B","C","D","G","H","I","K","M","N","R","S","T","V","W", "X" and "Y"
#'
#' TM_NN: check and return "A","C","G","I" and "T"
#'
#' @returns Return a sequence which fullfils the requirements of the given method.
#'
#' @author Junhui Li
#'
#' @references
#'
#' \code{citation("TmCalculator")}
#'
#' @examples
#'
#' ntseq <- c("ATCGBDHKMNRVYWSqq")
#' check_filter(ntseq,method='Tm_Wallace')
#' check_filter(ntseq,method='Tm_NN')
#'
#' @export check_filter
check_filter <- function(ntseq,method){
mySeq <- s2c(ntseq)
mySeq <- toupper(mySeq)
if (method == 'Tm_Wallace'){
baseset <- c("A","B","C","D","G","H","I","K","M","N","R","S","T","V","W","Y")
}else if (method == 'Tm_GC'){
baseset <- c("A","B","C","D","G","H","I","K","M","N","R","S","T","V","W","X","Y")
}else if(method == 'Tm_NN'){
baseset = c('A','C','G','T','I')
}else{
stop("Only methods 'Tm_Wallace' or 'Tm_GC' or 'Tm_NN' is allowed")
}
finalSeq <- NULL
#i='A'
for(i in mySeq){
if(i %in% baseset){
finalSeq <- append(finalSeq,i)
}
}
return(finalSeq)
}
Any scripts or data that you put into this service are public.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.