#!/usr/bin/perl -w

# Command line usage: perl<space>Integration.pl<space>original_sequence_file<space>direction_file<space>output_file
# Original_sequence_file is the output file produced by ESTNameConversion.pl.
# Direction_file is the file made from the Phrap output file with the following format:
#	Contigxxx
#	ESTxxxx.xx
#	ESTxxxx.xx
#	ESTxxxx.xx
# The funtion of this script is to integrate the sequences into the correct positions under each entry of the direction file.
# The output file looks like:
#	Contigxxx
#	>ESTxxxx.xx
#	sequence data
#	>EStxxxx.xx
#	sequence data
# The output file is ready for the standalone BLAST search against downloaded NCBI nr database.

my @Names;
my @NameLineNumber;
my @Seqs;
my $LineNumber = 1;

open (FILE1, "$ARGV[0]")
    or die "Cannot open first input file: $!";

while (<FILE1>) {
    chomp;
    
    if ($_ =~ /^>/) {
	push @Names, "$_";
	push @NameLineNumber, "$LineNumber";
	push @Seqs, "*";
    } else {
	push @Seqs, "$_";
    }
    $LineNumber += 1;
}

push @NameLineNumber, "$LineNumber";

close FILE1
    or die "Cannot close first input file: $!";

open (FILE2, "$ARGV[1]")
    or die "Cannot open second input file: $!";

my $Names = @Names;
my @ESTName;
my @Line;
my @SeqLength;
my @File2;

while (<FILE2>) {
    chomp;

    if ($_ =~ /^EST(\d+)/) {
	my $temp = $1;
	for (my $i = 0; $i < $Names; $i ++) {
	    if ($Names[$i] =~ /($temp)/){
		my $seqlength = $NameLineNumber[$i+1] - $NameLineNumber[$i] - 1;
		push @Line, "$NameLineNumber[$i]";
		push @SeqLength, "$seqlength";
		push @name, "$Names[$i]";
	    }
	}
    }
    push @File2, "$_";
}

open OUTPUT, ">$ARGV[2]"
    or die "Cannot open output file: $!";

my $j =0;

foreach (@File2) {
    if ($_ =~ /^EST/i) {
	print OUTPUT ">$name[$j]\n";
	for (my $k = $Line[$j]; $k < ($Line[$j] + $SeqLength[$j]); $k ++) {
	    print OUTPUT "$Seqs[$k]\n";
	}
	$j += 1;
    } else {
	print OUTPUT "$_\n";
    }
}

close OUTPUT
    or die "Cannot close output file: $!";