[Plan for finding homologies in nucleotide sequence databases using preliminarily calculated sequence samples].

A scheme of fast similarity search of nucleotide sequences is suggested based on sequence imaging, which results in chunks of information much less than original sequence but more specialized for comparison. Three methods were developed using three different imaging functions. The first is based on identity of local sites of up to twelve nucleotides, the second is based on statistical homology of local 42 nucleotide fragments, and the third is based on the homology of 100-150 nucleotide fragments and models the comparison of restriction maps. Each of them requires the library of sequence images. The total size of such a library is less than the size of sequences stored in compressed form. The sequences are aligned allowing local homology searches. The method reduces total time for a similarity search about 100-fold. The programs can be easily included in any software, which allows user to define his own set of sequences. One of the programs is implemented within DNA-SUN software and is used in Institute of Molecular Genetics and Institute of Molecular Biology.