Computational Toolset for Glycoconjugate Modeling and Simulation

Abstract

Characterizing glycoconjugates and protein-glycan interactions in the context of three-dimensional structures is important in understanding glycan's biological roles and develop efficient therapeutic agents. However, it is challenging to achieve this goal due to limitations in available experimental approaches and computational tools. To tackle this problem, we have developed GlycanStructure.ORG, a web portal to provide various computational tools. (1) Glycan Reader has been developed for automatic detection and annotation of carbohydrates, their chemical derivatives, and glycosidic linkages from PDB files (both in the PDB and mmCIF formats). Molecular system and input generations for the glycoconjugate simulation are available through CHARMM-GUI, http://www.charmm-gui.org. (2) The glycan fragment database (GFDB) provides an intuitive glycan sequence search tool that allows users to search for complex glycan structures in the PDB. After a glycan search, each selected glycosidic torsion angle distribution can be displayed and clustering analysis is used to provide most populated glycan structures, which can be useful for glycan structure modeling. (3) GS-align is an algorithm to identify the best structural alignment between glycan structure pairs and to provide a length-independent score of the structural similarity. This tool is useful to study the relationship between multiple glycoforms and biological functions, and glycan structure prediction. (4) Glycan Modeler is a template-based method to predict glycan structures from a given sequence using the PDB glycan structures as fragment template. After templates are identified by the random forest algorithm, Glycan Modeler provides two conformational sampling approaches that are assembly of rigid bodies and satisfaction of spatial restraints using by conformational space annealing (CSA).