#!/usr/bin/perl -w

#decided to have input file entered in command line
#call program followed by genome name. 
#the program assumes that a file with the extensions ptt and faa exist in the same dirctory. 
#####INPUT Name of multiple seq file containing ORF of genome, open file and assign IN filehandle ############# 
unless(@ARGV==1) {die "please provide genome name in command line \n  
file should contain multiple sequences in fasta format \n 
a file with the ptt table should be in the same directory\n\n";}
#$num=0;
$filename=$ARGV[0];  
@nameparts=split(/\./, $filename);
#print $parts[0];
$orfs="$nameparts[0]"."\.faa";
$ptt="$nameparts[0]"."\.ptt";

open(IN, "< $ptt") or die "cannot open $ptt:$!";
$line=<IN>; # read t1st line
	if ($line=~/complete genome/) {  #look forheader 
		print "$line\n";};
$line=<IN>; # read 2nd line
	print "$line\n";
$line=<IN>; # read 3rd line
if 	($line=~/Location	Strand/) {  #look for beginning of table 
 
	while (defined ($line=<IN>)){ # read through rest of table line by line
 
 		@parts=split/\t/,$line;
 		@fromto=split/\.\./,$parts[0];
 		$middle = (($fromto[1]+$fromto[0])/2);
 		print "$fromto[1]\t$fromto[0]\t$middle\t$parts[3]\t";
 $gi_hash{$parts[3]}=$middle;
 		print "\n";
	} 
}
@gi_names = sort(keys(%gi_hash));
$total=scalar(@gi_names);

print "total number of GIs= $total\n";
foreach (@gi_names) {
    print "gi number $_ is located at $gi_hash{$_}\n";
	}

close(IN);

# read in and process faa file

open(IN, "< $orfs") or die "cannot open $filename:$!";
$outfilename = "$nameparts[0]"."\.num\.faa";
open(OUT, "> $outfilename")||die "cannot open: $!";

##################################################


while (defined ($line=<IN>)){ # read through file line by line

	if ($line=~/^>/) {  #look for beginning of line starting with > (^ is an anchor for the beginning of the line)
 		$line =~ m/gi\|(\d+)\|/; #match gi|number capture number in $1
  		$num=$gi_hash{$1};
 		$line =~ s/^>//;
 		# print "$1 $num \n";
 		$line= ">"."$num\t"." $line";
		};
print "$line"; #print to screen 
print OUT "$line"; #print to OUT
	}

close(IN);
close(OUT);

#close(OUT);