We propose a motif discovery system that uses a modified PrefixSpan method to extract frequent patterns from an annotated sequence database that has such attributes as a sequence identifier (sequence-id), a sequence, and a set of items. The annotations are represented as the set of items in the database. Frequent sequence patterns and frequent item patterns are extracted from the annotated sequence database. Frequent sequence patterns are located in both identical and non-identical positions among those sequences. In general, the existing PrefixSpan method can extract a large number of identical patterns from the sequence databases. However, the method does not include a function to extract frequent patterns together with gaps or wild character symbols. This new method allows the incorporation of gap characters. Moreover, the method allows effective handling of the annotated sequence database that consists of a set of tuples including a sequence together with a set of items. Furthermore, the prototype has been applied to the evaluation of three sets of sequences that include the Zinc Finger, Cytochrome C, and Kringle motifs.