Accelerating Protein Sequence Search in a Heterogeneous Computing System