Wyślij emailem: Using homology relations within a database markedly boosts protein sequence similarity search