Signal processing approaches as novel tools for the clustering of N-acetyl-β-D-glucosaminidases

Nowadays, the clustering of proteins and enzymes in particular, are one of the most popular topics in bioinformatics. Increasing number of chitinase genes from different organisms and their sequences have beenidentified. So far, various mathematical algorithms for the clustering of chitinase genes have been used butmost of them seem to be confusing and sometimes insufficient. In the present study, as a first step, differentamino acids participating in panoply of chitinases, as a model protein, obtained from the NCBI GenBank,were digitized. Digitized data were normalized to the signal energy. Normalized data decomposed usingmother wavelet bior 5.5 to approximation (a1) and details (d1), at the first level. Corresponded coefficientshave been obtained and cross correlation between normalized, a1 and d1 coefficients of amino acidsequences were calculated. Maximum correlation was selected as similarity index and corresponded cladogramtrees were made. The results of this study showed that more optimal and reliable cladogram treecan be produced and better discrimination observed from d1 coefficients compared to normalizedsequences and opposed to a1 coefficients. Using suggested approach, the cladogram tree made from d1coefficients not only had more validity but also the drawback of the classic cladogram tree has been improved.