From 0f944dfba21c6bf9cf285935b87e7b96649523f5 Mon Sep 17 00:00:00 2001
From: Klaus Schliep <klaus.schliep@gmail.com>
Date: Tue, 17 Dec 2024 13:32:50 +0100
Subject: [PATCH] small improvement, add unalign

---
 R/phyDat_conversion.R | 22 ++++++++++++++++++++++
 R/pml_generics.R      |  7 +++++--
 man/as.phyDat.Rd      |  7 +++++++
 3 files changed, 34 insertions(+), 2 deletions(-)

diff --git a/R/phyDat_conversion.R b/R/phyDat_conversion.R
index d120e406..521af8c3 100644
--- a/R/phyDat_conversion.R
+++ b/R/phyDat_conversion.R
@@ -12,6 +12,10 @@
 #' \code{acgt2ry} converts a \code{phyDat} object of nucleotides into an binary
 #' ry-coded dataset.
 #'
+#' \code{unalign} converts a \code{phyDat} object of nucleotides or amino acids
+#' into a \code{DNAbin} or \code{AAbin} object in list form removing all gaps.
+#' These objects can be exported using \code{\link[ape]{write.FASTA}}.
+#'
 #' @aliases
 #' as.phyDat.character as.phyDat.data.frame as.phyDat.matrix
 #' as.MultipleAlignment as.MultipleAlignment.phyDat
@@ -367,3 +371,21 @@ acgt2ry <- function(obj){
                         levels = c("r", "y"), ambiguity = NULL)
   obj
 }
+
+
+#' @rdname as.phyDat
+#' @export
+unalign <- function(x){
+  if(inherits(x, "phyDat")){
+    if(attr(x, "type")=="DNA") x<- as.DNAbin(x)
+    if(attr(x, "type")=="AA") x<- as.AAbin(x)
+  }
+  stopifnot(inherits(x, "DNAbin") || inherits(x, "AAbin"))
+  x <- as.list(x)
+  for(i in 1:length(x)){
+    tmp <- as.character(x[[i]])
+    x[[i]] <- x[[i]][tmp!="-"]
+  }
+  x
+}
+
diff --git a/R/pml_generics.R b/R/pml_generics.R
index 4d93eb88..cde082e2 100644
--- a/R/pml_generics.R
+++ b/R/pml_generics.R
@@ -134,6 +134,8 @@ print.pml <- function(x, ...) {
 #' @param file a file name. File endings are added.
 #' @param save_rds logical, if TRUE saves the pml object as a rds file,
 #' otherwise the alignment is saved as a fasta file.
+## @param chi_sq logical, if TRUE performs $Chi^2$-test to check if sequences have similar
+## state composition.
 #' @param ... Further arguments passed to or from other methods.
 #' @returns \code{write.pml}  returns the input x invisibly.
 #' @seealso \code{\link{ancestral.pml}}, \code{\link{plotAnc}}
@@ -144,7 +146,7 @@ print.pml <- function(x, ...) {
 #' unlink(c("woodmouse.txt", "woodmouse_tree.nwk", "woodmouse_align.fasta"))
 #' @importFrom utils citation
 #' @export
-write.pml <- function(x, file="pml", save_rds=FALSE, ...){
+write.pml <- function(x, file="pml", save_rds=FALSE,...){
   digits <- -1
   if (hasArg("digits")) digits <- list(...)$digits
   write.tree(x$tree, file=paste0(file, "_tree.nwk"))
@@ -162,7 +164,8 @@ write.pml <- function(x, file="pml", save_rds=FALSE, ...){
   call$data <- quote(align)
   call$tree <- quote(tree)
   cat("tree <- read.tree(\"", file, "_tree.nwk\")\n", sep="")
-  cat("align <- read.phyDat(\"", file, "_align.fasta\", format=\"fasta\")",
+  type <- attr(x$data, "type")
+  cat("align <- read.phyDat(\"", file, "_align.fasta\", format=\"fasta\", type=\"", type,"\")",
       sep="")
   cat( "\nfit <- ")
   print(call)
diff --git a/man/as.phyDat.Rd b/man/as.phyDat.Rd
index d8e8a9d9..c721f3a5 100644
--- a/man/as.phyDat.Rd
+++ b/man/as.phyDat.Rd
@@ -25,6 +25,7 @@
 \alias{as.DNAbin.phyDat}
 \alias{as.AAbin.phyDat}
 \alias{genlight2phyDat}
+\alias{unalign}
 \title{Conversion among Sequence Formats}
 \usage{
 phyDat(data, type = "DNA", levels = NULL, return.index = TRUE, ...)
@@ -64,6 +65,8 @@ as.StringSet(x, ...)
 genlight2phyDat(x, ambiguity = NA)
 
 acgt2ry(obj)
+
+unalign(x)
 }
 \arguments{
 \item{data}{An object containing sequences.}
@@ -101,6 +104,10 @@ example for specifying "USER" defined data formats in the vignette
 
 \code{acgt2ry} converts a \code{phyDat} object of nucleotides into an binary
 ry-coded dataset.
+
+\code{unalign} converts a \code{phyDat} object of nucleotides or amino acids
+into a \code{DNAbin} or \code{AAbin} object in list form removing all gaps.
+These objects can be exported using \code{\link[ape]{write.FASTA}}.
 }
 \examples{