Nothing
#' Similarity score matrix
#'
#' Calculates the similarity score of all pairwise comparison from a list of
#' data frames.
#'
#' @param productive.seqs A list data frames of productive sequences generated
#' by the LymphoSeq function productiveSeq. "count" and "aminoAcid"
#' are a required columns.
#' @return A data frame of similarity scores calculated from all pairwise
#' comparisons. The similarity scores is a measure of the amount of overlap
#' between two samples. The value ranges from 0 to 1 where 1 indicates
#' the sequence frequencies are identical in the two samples and 0 indicates
#' no shared frequencies.
#' @seealso \code{\link{pairwisePlot}} for plotting results as a heat map.
#' @examples
#' file.path <- system.file("extdata", "TCRB_sequencing", package = "LymphoSeq")
#'
#' file.list <- readImmunoSeq(path = file.path)
#'
#' productive.aa <- productiveSeq(file.list = file.list, aggregate = "aminoAcid")
#'
#' similarityMatrix(productive.seqs = productive.aa)
#' @export
similarityMatrix <- function(productive.seqs) {
l <- length(productive.seqs)
m <- matrix(nrow = l, ncol = l)
rownames(m) <- names(productive.seqs)
colnames(m) <- names(productive.seqs)
for (i in 1:l) {
for (j in 1:i) {
m[i, j] <- similarityScore(productive.seqs[[i]], productive.seqs[[j]])
m[j, i] <- m[i, j]
}
}
m <- m[, colnames(m)[order(nchar(colnames(m)), colnames(m), decreasing = TRUE)]]
m <- m[colnames(m)[order(nchar(colnames(m)), colnames(m), decreasing = TRUE)], ]
d <- as.data.frame(m)
return(d)
}
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.