#!/usr/bin/perl

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

# rep_pos.pl
# Version 1

# Robert D. Cormia
# UCSC Programming for Bioinformatics III
# October 31, 2003

# Substantial help from Neil Cormia on this exercise

# Welcome to Intro. to Prog. for Bioinformatics III
# Class 3, Exercise 1

# 1. In this exercise we will work with the concept of counting aminoacids and
# noting down the respective positions. Write a Perl program called 'rep_pos'
# to count the aminoacids in a data file and note the positions as well. The
# data file looks like this:

# DFDKABLEGXKLRQDXVVMHMGQBA
# XCGGAYABSMKERDXGLQYASHDELHFFEET
# DKLZAZANXMKTIIDSVLGQTYTYQMETHERXXG
#
# Generate reports in the following fashion:
#
# Report I:
#
# Aminoacids: A
#   Pos 5: => 3
#   Pos 7: => 1
#   Pos 20: => 1
#   Pos 25: => 1
# Aminoacids: B
#   Pos 6: => 1
#   Pos 8: => 1
#   Pos 24: => 1
# ...
#
#Report II:
#
# At position: 1
#   D: => 2
#   P: => 1
#   X: => 1
# At position: 2
#   C: => 1
#   F: => 1
#   K: => 1
# At position: 3
#   D: => 1
#   G: => 1
#   L: => 1

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

use strict;

my $line;
my $count_aa  = 'count_aa.txt';
my $aa_output = 'aa_output.txt';
my @aa_string;
my $aa;
my %aa_hash;
my ( $i, $j );

print "\n";

open( INPUT_HANDLE, $count_aa )
  or die ("open() failed for 'fname'\nReason: $!\n\n");

open( OUTPUT_HANDLE, ">$aa_output" )
  or die ("open() failed for '$aa_output'\nReason: $!\n\n");

while ( $line = <INPUT_HANDLE> ) {

    print "$line \n";

    @aa_string = split ( //, $line );

    # i is the position within the string

    for ( $i = 0 ; $i < @aa_string, $aa = $aa_string[$i] ; $i++ ) {
        if ( !exists( $aa_hash{$aa} ) ) {
            $aa_hash{$aa} = &make_array_ref();
        }    # end if
        @{ $aa_hash{$aa} }[$i]++;

        # print "$aa_hash{$aa}";
    }

}    # end while

# create report

for ( $i = ord('A') ; $i <= ord('Z') ; $i++ ) {

    &report_aa( chr($i) );

}    # end for

# make a report subroutine

sub report_aa {

    if ( exists( $aa_hash{ $_[0] } ) ) {
        my @pos = @{ $aa_hash{ $_[0] } };

        print OUTPUT_HANDLE "Aminoacid: $_[0]\n";
        print "Aminoacid: $_[0]\n";

        for ( $j = 0 ; $j < @pos ; $j++ ) {

            if ( $pos[$j] > 0 ) {
                print OUTPUT_HANDLE "Pos $j: => $pos[$j]\n";
                print "Pos $j: => $pos[$j]\n";
            }

        }
        print "\n";
        print OUTPUT_HANDLE "\n";
    }
}    # end subroutine

# make_array subroutine creates a reference to an array

sub make_array_ref {
    my @array;
    return \@array;
}

    __END__