Local Patterns in Biological Sequences

Many recent advances in the field of biology are due to the widespread and ever-increasing availability of sequence data, both DNA and protein. Local Patterns in Biological Sequences presents three different algorithms that make use of this wealth of information: CYRCA, HexDiff, and PreCUSA. CYRCA was designed to detect weak sequence similarity between protein families using conserved protein sequence patterns and graph theory. HexDiff used frequencies of DNA patterns to detect a specific type of regulatory sequence called cis-regulatory modules. The third approach, drawing on the ideas explored in CYRCA and HexDiff, used structural attributes predicted from DNA sequence as an alternative source of information for detecting regulatory sequences in the genome. The common theme for these three algorithms is extracting local features from long biological sequences and utilizing them to predict large-scale sequence properties.