No you don't fill @F. The -a switch does it for you. see perldoc perlrun The code above is complete script. Just run it as shown in usage.

Now explanations:

#!/usr/bin/perl -lan
# Usage: thisscript.pl genes.txt


# %seen is a hash where we store keys composed from seen rows
# @F is an array of 2 elements $F[0] is a first column of your file an
+d $F[1] is the second one (see [doc://perlrun] -a switch
# So the key for row is composed by concatenation of sorted columns

if ( $seen{ join ' ', sort @F }++ ) {

# %uniq hash will keep seen columns first and second

    $uniq{$F[0]}++;  
    $uniq{$F[1]}++;
}
# the code above is loops over each row of file because of -n switch s
+ee [doc://perlrun]

END {  # this block run just before exit
    print for keys %uniq;
}
[download]