One popular and reasonably effective approach is to shuffle the order of amino acids in each protein encoded by the genome, generating a database of predicted spectra from these shuffled sequences.