#!/usr/bin/perl

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

# rep_pos_hash.pl
# Version 1

# Robert D. Cormia
# UCSC Programming for Bioinformatics III
# November 7, 2003

# Substantial help from Neil Cormia and Son Nguyen
# on this exercise, mostly in creating reports 1 & 2.

# 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

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

# first read in sequences into a temp array
# create two hashes by reading in arrary reference
# create two print reports acting on the hashes
# first print report sorts hash by letter
# second print report sorts hash by number
# liberal use of print blank lines to seperate values

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

# Declare all variables

use strict;

my $line;
my $count_aa  = 'count_aa.txt';
my $aa_output = 'aa_output.txt';
my @aa_array;
my %hash_report_1;
my %hash_report_2;
my $i;

# Open input file to read in sequences of amino acids

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

# Open output file to read out the two reports

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

# Print sequences as read in

print "Sequences: \n\n";
print OUTPUT_FILE "Sequences: \n\n";

while ( $line = <INPUT_FILE> ) {

    next if ( $line =~ /^$/ );    # skip if blank line

    chomp($line);                 # remove carriage return
    print "$line \n";             # space before reports
    print OUTPUT_FILE "$line \n";

    @aa_array = split ( '', $line );    # read into temp array
    my $array_len = @aa_array;          # determine array length

    # i is the position within the string
    for ( $i = 0 ; $i < $array_len ; $i++ ) {

        # do report 1 and 2 in a for each loop
        # pass arrary references into each hash

        $hash_report_1{ $aa_array[$i] }{$i} += 1;
        $hash_report_2{$i}{ $aa_array[$i] } += 1;

    }

}    # End while

# Call print_report_1 to create report 1

&print_report_1( \%hash_report_1 );

# Call print_report_2 to create report 2

&print_report_2( \%hash_report_2 );

#
# Print reports using subroutines written by Son Nguyen
#

##############  Subroutine Report 1  ##############

sub print_report_1 {
    my %hash = %{ $_[0] };
    my $amino_acid;
    my $pos;

    # Print blank line and Report I header

    print "\n";
    print OUTPUT_FILE "\n";
    print "Report I:\n\n";
    print OUTPUT_FILE "Report I:\n\n";

    foreach $amino_acid ( sort keys %hash ) {

        print "\n";
        print OUTPUT_FILE "\n";
        print "Amino Acid:  $amino_acid\n";
        print OUTPUT_FILE "Amino Acid:  $amino_acid\n";

        # Foreach loop to sort keys by amino acid by letter
        # sort { $a <=> $b } $a and $b have values -1, 0, +1
        # creates a sorted list by letter in amino acid array

        foreach $pos ( sort { $a <=> $b } keys %{ $hash{$amino_acid} } )
        {    #start counting position from 1
            my $pos_count = $pos + 1;
            print "Pos $pos_count :=>  $hash{$amino_acid}{$pos}\n";
            print OUTPUT_FILE "Pos $pos_count :=>  $hash{$amino_acid}{$pos}\n";
        }
    }
}

##############  Subroutine Report 2  ##############

sub print_report_2 {
    my %hash = %{ $_[0] };
    my $amino_acid;
    my $pos;

    # Print blank line and Report II header

    print "\n";
    print OUTPUT_FILE "\n";
    print "Report II:\n\n";
    print OUTPUT_FILE "Report II:\n\n";

    # Foreach loop to sort keys by amino acid by number
    # sort { $a <=> $b } $a and $b have values -1, 0, +1
    # creates a sorted list by position in amino acid array

    foreach $pos ( sort { $a <=> $b } keys %hash ) {

        print "\n";
        print OUTPUT_FILE "\n";
        my $pos_count = $pos + 1;
        print "At position:  $pos_count\n";
        print OUTPUT_FILE "At position:  $pos_count\n";

        # sort { $a <=> $b } $a and $b have values -1, 0, +1
        # sort base on alpha numerical values

        foreach $amino_acid ( keys %{ $hash{$pos} } )
        {    #start counting position from 1
            print "$amino_acid :=>  $hash{$pos}{$amino_acid}\n";
            print OUTPUT_FILE "$amino_acid :=>  $hash{$pos}{$amino_acid}\n";
        }
    }
}

    __END__