#!/usr/bin/perl -w
#
# cure_cycle_split.pl
# A simple Wrapper script to perform n cycles of reference sequence "curing", calling the script cure_refseqs.pl
#
# Janet.Higgins@bbsrc.ac.uk,17/2/2011

# Script requires the naive reference sequence in both fasta and bfa format, together with the reads
# in bfq format obtained from a diploid progenitor of the polyploid genome.
# The reads have been split into two parts,each containing 40bp reads to improve alignment efficiency.
# mapmerge is used to merge the maps from the two sets of reads.
# It assumes the MAQ program is in the user's path.
# The default number of cycles in four, in our application we found six cycles were required.

unless (@ARGV >= 4) {
  print "Usage: cure_cycle_split.pl <starting ref.fa> <starting ref.bfa> <reads1.bfq> <reads2.bfq> <cycles[default 4]>\n";
  exit;
}

my $cycles = $ARGV[4] || 4;

# Open a log file for combined STDOUT/STDERR from Maq
my $logfile = "curing.$$.log";

# We have to allow file clobbering, simply on disc space grounds
# First we copy the original refseq files so that they can be clobbered safely
system ("cp $ARGV[0] current.fa");
system ("cp $ARGV[1] current.bfa");

for my $i (1 .. $cycles) {
  print "Starting curing cycle $i/$cycles ...\n";
  print "Mapping ($i/$cycles cycles)... \n";
  system ("maq map current1.map current.bfa $ARGV[2] >> $logfile 2>&1 ");
  system ("maq map current2.map current.bfa $ARGV[3] >> $logfile 2>&1 ");
  print "Merging maps ($i/$cycles cycles)... \n";
  system ("maq mapmerge current.map current1.map current2.map >> $logfile 2>&1");
  print "Assembling consensus ($i/$cycles cycles)... \n";
  system ("maq assemble current.cns current.bfa current.map >> $logfile 2>&1");
  print "Extracting consensus ($i/$cycles cycles)... \n";
  system("maq cns2fq current.cns > current.cns.fastq");
  print "Refseq curing ($i/$cycles cycles)... \n";
  system ("cure_refseqs.pl current.fa current.cns.fastq cured.fa");
  print "Converting cured refseq ($i/$cycles cycles)... \n";
  system ("maq fasta2bfa cured.fa cured.bfa");
  print "Rotating files ($i/$cycles cycles)\n";
  system ("mv cured.fa current.fa");
  system ("mv cured.bfa current.bfa");
}

# Extract salient details from logfile, print to STDOUT and then delete
print "\nMapping efficiencies over curing cycles\n";
open (LOG, "<$logfile") or die "Couldn't open logfile for parsing ($!)\n";
while (<LOG>) {
  if (/^\[match_data2/ || /^-- \(total/) { print $_; }
}

unlink $logfile;