[33] Analysis of compositionally biased regions in sequence databases

This chapter discusses the analysis of compositionally biased region in sequence databases. Programs sequence (SEG) and protein sequence (PSEG) are tuned for amino acid sequences and nucleotide sequence (NSEG) for nucleotide sequences. The programs can be applied to either individual sequences, including whole chromosomes if appropriate, or entire sequence databases. Compositional complexity is based only on residue composition, regardless of the patterns or periodicity of sequence repetitiveness. This contrasts with some alternative methods that use counts of k-grams to define residue patterns and clustering. Complexity, pattern, and periodicity are distinct abstract attributes of simple sequences. For genomic studies, it is essential to view compositional bias in the context of many types of other features, such as recognizable functional sites, transcripts, coding sequences, and homologies. For this purpose, the SEG family of programs is being integrated into software packages, or workbenches, that have graphic multilevel browsing facilities and include zoom functions.