R/orderContigsTSP.R

In oneillkza/ContiBAIT: Improves Early Build Genome Assemblies using Strand-Seq Data

#####################################################################################
#' Attempt to order contigs within linkage groups using travelling salesperson algorithm
#' @useDynLib contiBAIT
#' @import Rcpp TSP
#' @import TSP
# 
#' @param linkageGroupReadTable dataframe of strand calls (product of combineZeroDists or preprocessStrandTable)
#' @param reps Number of repetitions for he 2-opt TSP solver algorithm
#' @return list of two members: 1) contig names in order, 2) the original data.frame entered into function correctly ordered 
########################################################################################

orderContigsTSP <- function(linkageGroupReadTable, reps)
{

  factorisedLGReadTable <- data.frame(apply(linkageGroupReadTable, 2, as.factor), stringsAsFactors = TRUE)
  
  dists <- as.matrix(daisy(factorisedLGReadTable))
  
  # Add a dummy "city" equidistant to all others:
  dists <- cbind(dists, rep(1, nrow(dists)))
  dists <- rbind(dists, rep(1, ncol(dists)))
  rownames(dists)[nrow(dists)] <- 'dummy'
  colnames(dists)[nrow(dists)] <- 'dummy'
  
  #Compute optimal TSP route, and remove dummy:
  contig.tsp <- TSP(dists)
  contigsOrder <- solve_TSP(contig.tsp, method='2-opt', control=list(rep=reps))
  contigsOrder <- labels(contigsOrder)[-c(which(labels(contigsOrder)=='dummy'))]

  return(list(orderVector=contigsOrder, orderedMatrix= factorisedLGReadTable[contigsOrder,]))
}