First of all, you have a mistake in your code. The first line in your loop should be:

  my @split_hit_total = split(/\s+/, $_);
[download]

since $_ is your loop variable here, not $hit_total. (use strict is good for catching things like that, BTW!)

That said, to answer your question: repurpose your HoA so that it stores (for each $protein_ac) the lowest encountered iEvalue along with its corresponding range, rather than all ranges. E.g. instead of pushing the range, do this:

  if(!defined($HoA_total_hits{$protein_ac}) || $HoA_total_hits{$protei
+n_ac}->{'iEvalue'} > $iEvalue) {
    $HoA_total_hits{$protein_ac}->{'iEvalue'} = $iEvalue;
    $HoA_total_hits{$protein_ac}->{'range_b'} = $range_b;
  }
[download]

Untested in the absence of example data... and I've not had my morning brew yet, so apologies if this is obvious rubbish.

(Side note -- I turned your HoA into a HoH here, but didn't change the name. I'd advise against putting the type of a variable in its name anyway.)

EDIT: using the sample data from Re^2: How can I add an entry in a has of arrays based on numeric comparison?, here's what Data::Dumper has to say about the contents of $HoA_total_hits after running this:

$VAR1 = {
          'I5EU07' => {
                        'range_b' => '232-824',
                        'iEvalue' => '3.4e-137'
                      }
        };
[download]

Does that look right?

Aha, thanks!
So, I must change the structure to HoH instead of HoA to achieve the desired result?
(All these advanced structures seem so complicated to me!!!!)

You don't have to; you can continue using a HoA as well, replacing (say) ->{'iEvalue'} with ->[0] and ->{'range_b'} with ->1. If you did this, the result for your sample data would be:

$VAR1 = {
          'I5EU07' => [
                        '3.4e-137',
                        '232-824'
                      ]
        };
[download]

It's pretty much the same thing, but I personally prefer using hashes over arrays unless my "keys" are naturally numeric; you have to type a bit more, but your code'll be self-documenting. It's always obvious what ->{'iEvalue'} means, even years later when you're reading someone else's code, but the same is not true for ->[0].

Can you provide some sample data, and a corresponding example of your desired output?

Hi,
so, some sample lines could be:


STN              PF07660      597 I5EU07               -           125
+9    1e-140  467.1  11.7   1   2  5.9e-140  3.4e-137  455.5   7.3    
+ 9   595   238   822   232   824 0.97 -
STN              PF07660      597 I5EU07               -           125
+9    1e-140  467.1  11.7   2   2   3.9e-05     0.023   10.2   0.0   1
+48   227   864   949   842   962 0.80 -
[download]


and, in that case, we would need to store the first line, because the $iEvalue for the first line is 3.4e-137, which is lower (thus better) than the second line's $iEvalue which is 0.023.
So, the only thing I need is somehow, if I have the same $protein_ac (in this example I5EU07) and more than one lines of the same $pfam_ac_hit (in this example PF07660), I keep only the one with the lowest $iEvalue.