G'day davi54,

"I want to count the number of alphabets in each entry, length of each entry, etc."

That's confusing as you combine the counts for all entries in @count. Other variables, outside the while loop, appear misplaced as they look like they're associated with individual entries: I'd expect them to be inside the while loop.

The following code collects all the data that I believe you want. You can combine values for all entries if necessary.

#!/usr/bin/env perl

use strict;
use warnings;

my %results;

{
    local $/ = '';

    while (my $record = <DATA>) {
        $record =~ s/\A>(.+?)$//m;
        my $entry = $1;
        $record =~ s/\s//gm;
        $results{$entry}{len} = length $record;

        for (0 .. $results{$entry}{len} - 1) {
            ++$results{$entry}{count}{substr $record, $_, 1};
        }
    }
}

use Data::Dump;
dd \%results;

__DATA__
>sp_0005
VQLQESGGGLVQAGGSLRLSCAASGRAVSMYNMGWFRQAPGQERELVAAISRGGSIYYA
DSVKGRFTISRDNAKNTLYLQMNNLKPEDTGVYQCRQGSTLGQGTQVTVSS

>sp_0017
HVQLVESGGGSVQAGGSLRLTCAASGFTFSNYYMSWVRQAPGKGLEWVSSIYSVGSNGYY
ADSVKGRSTISRDNAKNTLYLQMNSLKPEDTAVYYCAAEPGGSWWDAYSYWGQGTQVTVS S
[download]

Extract of output:

{
  sp_0005 => {
               count => {
                 A => 9,
                 C => 2,
                 ...,
                 W => 1,
                 Y => 5,
               },
               len => 110,
             },
  sp_0017 => {
               count => {
                 A => 10,
                 C => 2,
                 ...,
                 W => 5,
                 Y => 10,
               },
               len => 121,
             },
}
[download]

Notes:

• You posted sample data but did so using paragraph text. Please use <code> tags for this sort of thing. I've made a best guess at what the original looked like: I was unsure about the embedded space at the end of the second entry (cf. 'QVTVSS' in 1st entry and 'QVTVS S' in 2nd entry) so left it as written.
• I've truncated the names to avoid excessive text wrapping. They're still unique but now just look like sp_NNNN.
• When you modify special variables, such as $/ here, you should localise the change in the smallest scope possible (see my code for an example of this).
• I've used substr instead of a regex to get the counts. Perl's string-handling functions are typically faster than regexes that achieve the same outcome.
• Do note that this code is only intended to show a technique. It does not contain any I/O or formatted reporting.