G'day rnaeye,

I'm a tad late to the party on this one: I've been indisposed and haven't logged in for about a week.

What follows is a series of techniques; some or all may be useful for you. I've basically dealt with this in two phases: the extraction of the data and the reporting of results. Both of these take into consideration your desire to deal with the data in chunks and output as a single file or with multiple files (one per chunk).

I'll show the code first. Notes follow which explain what I did differently and why.

#!/usr/bin/env perl

use 5.28.0;
use warnings;
use autodie;

use Text::CSV;

use constant {
    SEP_CHAR => "\t",
    BED_CHR => 0,
    BED_START => 1,
    BED_END => 2,
    BED_STRAND => 5,
    BED_LINE => 6,
};

my (%start_plus, %start_minus, %start_dud);

my %dispatch = (
    '+' => sub { ++$start_plus{$_[BED_CHR]}{$_[BED_START]} },
    '-' => sub { ++$start_minus{$_[BED_CHR]}{$_[BED_END]} },
    '0' => sub { $start_dud{$_[BED_CHR]}{$_[BED_LINE]} = $_[BED_STRAND
+] },
);

my $csv = Text::CSV::->new({sep_char => SEP_CHAR});

while (my $row = $csv->getline(\*DATA)) {
    next if 0 == index $row->[0], '#';
    my $sign = exists $dispatch{$row->[BED_STRAND]} ? $row->[BED_STRAN
+D] : 0;
    $dispatch{$sign}->(@$row, $.);
}

# This part for testing and demonstration purposes only
use Data::Dump;
say '=' x 10, ' Raw Extracted Data ', '=' x 20;
say 'Plus Strands';
say '-' x 50;
dd \%start_plus;
say '-' x 50;
say 'Minus Strands';
say '-' x 50;
dd \%start_minus;
say '-' x 50;
say 'Problem Strands';
say '-' x 50;
dd \%start_dud;
say '=' x 50;
# End this testing & demo part

my %filehandle_for;

for my $chr (sort keys %start_plus) {
    for my $file_id (ALL => $chr) {
        my $fh = _get_results_fh(\%filehandle_for, $file_id);
        $fh->say(join SEP_CHAR, CHR => 'Plus Strand BedGraph:');

        for my $key (sort {$a <=> $b} keys %{$start_plus{$chr}}) {
            $fh->say(join SEP_CHAR, $chr, $key + 1, $start_plus{$chr}{
+$key});
        }
    }
}

# This part for testing and demonstration purposes only
say "\n", '=' x 10, ' Output Files ', '=' x 26;
for my $outfile (qw{
    pm_11135183_bio_bed_modified_chr7_results.bed
    pm_11135183_bio_bed_modified_chr8_results.bed
    pm_11135183_bio_bed_modified_ALL_results.bed
}) {
    say $outfile;
    say '-' x 50;
    system cat => $outfile;
    say '-' x 50;
}
# End this testing & demo part

sub _get_results_fh {
    my ($fh_for, $chr) = @_;

    return $fh_for->{$chr} || _gen_results_fh($fh_for, $chr);
}

sub _gen_results_fh {
    my ($fh_for, $chr) = @_;

    my $fmt = 'pm_11135183_bio_bed_modified_%s_results.bed';
    my $filename = sprintf $fmt, $chr;
    open my $fh, '>', $filename;
    $fh_for->{$chr} = $fh;

    return $fh;
}

__DATA__
#chromosome start   end position    col-5   strand
chr7    127471196   127472363   Pos1    0   @
chr7    127471196   127472363   Pos1    0   +
chr7    127471196   127472363   Pos1    0   +
chr7    127471196   127472363   Pos1    0   +
chr7    127471196   127472363   Pos1    0   +
chr7    127471196   127472363   Pos1    0   +
chr7    127472363   127473530   Pos2    0   +
chr7    127473530   127474697   Pos3    0   -
chr7    127474697   127475864   Pos4    0   -
chr7    127474697   127475864   Pos4    0   _
chr8    40000   40500   Pos5    0   +
chr8    40000   40600   Pos5    0   +
chr8    40000   40650   Pos5    0   =
chr8    40000   40650   Pos5    0   +
chr8    41000   41200   Pos6    0   -
[download]

Here's the output from a sample run:

$ ./pm_11135183_bio_bed_modified.pl
========== Raw Extracted Data ====================
Plus Strands
--------------------------------------------------
{ chr7 => { 127471196 => 5, 127472363 => 1 }, chr8 => { 40000 => 3 } }
--------------------------------------------------
Minus Strands
--------------------------------------------------
{ chr7 => { 127474697 => 1, 127475864 => 1 }, chr8 => { 41200 => 1 } }
--------------------------------------------------
Problem Strands
--------------------------------------------------
{ chr7 => { 2 => "\@", 11 => "_" }, chr8 => { 14 => "=" } }
==================================================

========== Output Files ==========================
pm_11135183_bio_bed_modified_chr7_results.bed
--------------------------------------------------
CHR     Plus Strand BedGraph:
chr7    127471197       5
chr7    127472364       1
--------------------------------------------------
pm_11135183_bio_bed_modified_chr8_results.bed
--------------------------------------------------
CHR     Plus Strand BedGraph:
chr8    40001   3
--------------------------------------------------
pm_11135183_bio_bed_modified_ALL_results.bed
--------------------------------------------------
CHR     Plus Strand BedGraph:
chr7    127471197       5
chr7    127472364       1
CHR     Plus Strand BedGraph:
chr8    40001   3
--------------------------------------------------
[download]

Notes:

• I've used the warnings pragma and strongly recommend you do the same in all Perl programs.
  • The "use 5.28.0;" automatically gives you the strict pragma. In the absence of a "use VERSION;" (>=5.12) I'd also strongly recommend explicitly including strict; of course, you may already know this.
• I've used the autodie pragma. This takes the tedium out of hand-crafting I/O exception messages; it's also far less error prone — note that your message, "Cannot open file." doesn't tell the user what file had that problem (autodie will do that for you).
• Next, you'll see Text::CSV. Although the CSV abbreviation stands for "comma-separated values", this module can be used with other separator characters. Notice that I used the constant SEP_CHAR; if the separator ever changes, there's only one thing to change in the code to accommodate such a change. Text::CSV allows this with "{sep_char => SEP_CHAR}". One final point, if you also have Text::CSV_XS installed it will run much faster (but that's not a requirement).
• I've included a dispatch table (%dispatch) to handle your processing generically. Notice a little further down that my while loop iterating DATA only needs three statements.
• I've shown the data extracted. See Data::Dump if you're unfamiliar with that module.
• To handle the potential output to multiple files, I've used a hash (%filehandle_for) to cache the filehandles needed. They only need to be opened once; only those that are needed are opened. The code is fairly straightforward: _get_results_fh() will return a cached filehandle or, if it hasn't been created yet, will get a new one via _gen_results_fh().
  • While writing these notes, it occurs to me that this could be improved such that the hashref doesn't need to be passed with every invocation. Possibly something like this which keeps the hash private to the two routines that need it and simplifies all of the calling code:

    {
        my %filehandle_for;
    
        sub _get_results_fh { my ($chr) = @_; ... }
    
        sub _gen_results_fh { my ($chr) = @_; ... }
    }
    [download]

• You may have noticed that I've used "$fh->say(...)" instead of "say $fh (...)". See "perlobj: Indirect Object Syntax" for more about that.
• You didn't provide any indications about how the output should look when split across multiple files. I just made some guesses and provided alternatives: I expect you'll want to make changes there. If you want to give users choices about the output, take a look at the core module, Getopt::Long.
• I added a few extra lines to __DATA__ for testing purposes. Some represent possible typos, e.g. '_' instead of '-'; consider doing this sort of thing in all of your code.