Development of a new glycan score matrix

Glycans are chains of monosaccharides also known as oligosaccharides. Since glycans consist of monosaccharides having multiple hydroxyl groups which bind with potentially multiple other monosaccharides, glycans have very complicated structures compared to nucleic acid or protein sequences. The complexity is complicated further by various glycosidic linkage patterns which vary according to anomeric configuration as well [1]. Possibly because of this structural complexity, glycans are extremely vital for the development and functioning of multi-cellular organisms. They are recognized by a variety of proteins and consequently promote or inhibit various diseases. Despite all the biological knowledge accumulated over the years in glycobiology, the informatics of glycan structures and glycome informatics are still at an early stage. Amino acid score matrices reflect the physico-chemical similarity between pairs of amino acids according to the evolutionary relationships between protein sequences. By aligning amino acid sequences based on appropriate amino acid score matrices, highly accurate alignments can be obtained. Although glycan score matrices are different from protein score matrices, appropriate glycan score matrices help to improve the accuracy of glycan structure alignment. It is surmised, furthermore, that physico-chemical properties of glycosidic linkages and monosaccharides may be obtained from such matrices. Therefore, this research focuses on generating and analyzing new glycan score matrices based on previous work which illuminated the potential of glycan matrices [2]. In previous work, it was understood that the data registered in most glycan databases are biased towards N-glycans due to the relatively larger amount of research having been focused on these structures. For instance, the number of N-glycans registered in KEGG is over 2000, while the number of glycerolipid entries is about 20. When a glycan score matrix is calculated using these glycan data, it is inevitable that the number of times by which N-glycan is aligned will be more than the number of times by which a glycerolipid is aligned. This was not taken into consideration previously. Thus, to solve this problem, we modified our matrix generation algorithm.