Identification of Sequence Patterns, Motifs and Domains

The evolutionary process is constrained by the structure and function of biological molecules. These constraints appear in macromolecular sequences as conserved regions. Because of the tight coupling between sequence conservation and molecular function, a huge amount of effort has gone into developing methods for describing sequence patterns, motifs, and domains, and into developing efficient methods for identifying novel examples of known motifs. In computational biology, the terms patterns, motifs, and domains are used somewhat interchangeably, depending on the size of the molecular patter under discussion; in general shorter patterns are often referred to as motifs, signatures, or patterns, while longer patterns are likely to be referred to as domains. Current approaches have their roots in sequence alignment, but more recent methods employ many techniques from machine learning.