A fixed-point alignment analysis technique is presented which is designed to locate common sequence motifs in collections of proteins or nucleic acids. Initially a program aligns a collection of sequences by a common sequence pattern or known biological feature. The common pattern or feature (fixed-point) may be a user-specified sequence string or a known sequence position like mRNA start site, which may be taken directly from the annotated feature table of GenBank. Once all alignment markers are located, the sequences are scanned for occurrences of given oligomers within a specified span both upstream and downstream of the fixed-point. The occurrences may then be plotted as a function of the position relative to the fixed-point, displayed as an actual sequence alignment or selectively summarized via various program options. Applications of the technique are discussed.

We present an algorithm to detect distances between oligonucleotides in large collections of nucleic acids sequences. The ratios of actual frequencies of occurrence of short oligonucleotides at a given distance to the corresponding expected frequencies were analyzed in four categories of DNA sequences leukaryotic exons, bacterial genes, introns and non-Alu repeated DNAs). Three base periodic occurrences (independent of the reading frame) of all combinations of mononucleotides and repeats of all dinucleotides was characteristic for protein coding regions. This was also the case with the majority of trinucleotides (including translational stop signals) in these regions. Mirror-symmetric trinucleotides (except GCG and CGC) displayed a strong tendency to be two base periodically repeated in introns. Some two and three base periodic motifs were also observed in repeated DNAs. The possible biological implications of outstanding three base periodicities in bacterial genes and eukaryotic exons are discussed.