If the order of lines isn't important, then you should use hashes:
Update: refactored to use one hash and to print only the first column.

my %seen;
#open DATA1 here
++$seen{$_} while (<DATA1>);
#close DATA1 here
#open DATA2 here
while (<DATA2>) {
    # Print any lines that are found, that weren't in DATA1
    print((split)[0], "\n") unless (defined(delete $seen{$_}));
}
# print what's left
print((split)[0], "\n") for (keys %seen);
[download]

If this quick and dirty approach isn't what you're looking for, check out the Algorithm::Diff module.