A rapid access motif database (RAMdb) with a search algorithm for the retrieval patterns in nucleic acids or protein databanks.

We present here a codification structure, entirely interfaced with the main packages for biomolecule database management, associated with a new search algorithm to retrieve quickly a sequence in a database. This system is derived from a method previously proposed for homology search in databanks with a preprocessed codification of an entire database in which all the overlapping subsequences of a specific length in a sequence were converted into a code and stored in a hash-coding file. This new algorithm is designed for an improved use of the codification. It is based on the recognition of the rarest strings which characterize the query sequence and the intersection of sorted lists read in the codification structure. The system is applicable to both nucleic acid and protein sequences and is used to find patterns in databanks or large sets of sequences. A few examples of applications are given. In addition, the comparison of our method with existing ones shows that this new approach speeds up the search for query patterns in large data sets.