Computationally Efficient Cluster Representation in Molecular Sequence Megaclassification