Fig. 2

Addition of restricted linkage nodes improves selection of candidate motifs for glycan binding data. In this illustrative example, there is a single glycan (Gal β1-3GalNAc) capable of binding to a candidate lectin (e.g. PNA), while sialylation of the galactose residue (Neu5Ac α2-3Gal β1-3GalNAc and Neu5Ac α2-6Gal β1-3GalNAc) restricts binding. Generation of subtrees from these three glycans yields a set of potential motifs that could be used to discriminate between binders and non-binders. Note that one of these subtrees contains a ’restricted linkage’ node, to indicate the absence of a connection at positions 3 and 6 on the terminal galactose; there are connections at these positions within the non-binding set. This restricted linkage node is indicated by an X. Without consideration of restricted linkage nodes, there are no subtrees that are unique to the binding set. However, with addition of restricted linkage nodes, there is a single subtree from the binding set that adequately discriminates between binding and non-binding glycans. This candidate motif is marked with an asterisk. All glycan motif structures are shown in SNFG [51], modified with restricted linkages. Each restricted linkage, with corresponding carbon numbers, terminates in a cross in place of a residue symbol, according to the key