use strict;
use warnings;
use Data::Dumper;

my %sequences;    # This hash will hold sequence ID's and the
                  # associated longest values.

while( defined( my $id = <DATA> ) ) {    # Read the ID from file.
    defined( my $sequence = <DATA> )     # Read the sequence.
        or die "Odd number of lines in DATA.\n";
    chomp for ( $id, $sequence );        # Chomp the input.
    $id =~ s/^>//;                       # Strip the > character
                                         # from the ID.
    if(
        exists( $sequences{ $id } ) and 
        length( $sequences{ $id } ) >= length( $sequence ) 
    ) {
        # Current sequence not longer. Skip to next record.
        next;
    } else {
        # Current sequence is longer.  Keep track of it.
        $sequences{ $id } = $sequence;
    }
}

#  Now %sequences contains all the longest strings for each ID.
#  Print the hash...
print Dumper \%sequences;

__DATA__
>protein1
ASFGTHTRHTHRHTHTRHTRHTR
>protein2
ERYRYTRYHTRHTGEFEWWFEEFFFFREFRGRE
>protein3
AWEERERGRGRGREGRGREGRRRRRRRRTTHTHTRHRHTRHTR
>protein2
AASEFEFEFE
>protein4
REYTRHTRGRVEVCREVR