The GlycanBuilder: a fast, intuitive and flexible software tool for building and displaying glycan structures

Abstract

Background: Carbohydrates play a critical role in human diseases and their potential utility as biomarkers for pathological conditions is a major driver for characterization of the glycome. However, the additional complexity of glycans compared to proteins and nucleic acids has slowed the advancement of glycomics in comparison to genomics and proteomics. The branched nature of carbohydrates, the great diversity of their constituents and the numerous alternative symbolic notations, make the input and display of glycans not as straightforward as for example the amino-acid sequence of a protein. Every glycoinformatic tool providing a user interface would benefit from a fast, intuitive, appealing mechanism for input and output of glycan structures in a computer readable format.

Results: A software tool for building and displaying glycan structures using a chosen symbolic notation is described here. The "GlycanBuilder" uses an automatic rendering algorithm to draw the saccharide symbols and to place them on the drawing board. The information about the symbolic notation is derived from a configurable graphical model as a set of rules governing the aspect and placement of residues and linkages. The algorithm is able to represent a structure using only few traversals of the tree and is inherently fast. The tool uses an XML format for import and export of encoded structures.

Conclusion: The rendering algorithm described here is able to produce high-quality representations of glycan structures in a chosen symbolic notation. The automated rendering process enables the "GlycanBuilder" to be used both as a user-independent component for displaying glycans and as an easy-to-use drawing tool. The "GlycanBuilder" can be integrated in web pages as a Java applet for the visual editing of glycans. The same component is available as a web service to render an encoded structure into a graphical format. Finally, the "GlycanBuilder" can be integrated into other applications to create intuitive and appealing user interfaces: an example is the "GlycoWorkbench", a software tool for assisted annotation of glycan mass spectra. The "GlycanBuilder" represent a flexible, reliable and efficient solution to the problem of input and output of glycan structures in any glycomic tool or database.

Figure 1

Symbolic representation of monosaccharides. Representation of monosaccharides with geometric shapes as described in the notations used by the Consortium for Functional Glycomics (CFG) and the Oxford Glycobiology Institute (UOXF).

Figure 2

Orientation and style of linkages in UOXF notation. In the symbolic notation adopted by the Oxford Glycobiology Institute the anomeric state is represented by varying the style of the line (dashed or continuous) while the linkage position is represented by the orientation of the saccharide in respect to its parent.

Figure 3

Example of a structure represented using CFG and UOXF notations. The same structure is drawn here using four different symbolic notations: a) the CFG standard notation, b) the CFG black and white notation, c) the CFG hybrid notation with placement of residues in respect to linkage position, and d) the UOXF notation. These drawings have been created using the "GlycanBuilder".

Figure 4

Available positions for a residue around its parent. During the positioning phase of the rendering algorithm each residue is placed in a generic spatial region around its parent. These regions are used during the computation of bounding boxes to decide the alignment and spatial placement of each residue's children. The position is identified by the relative angle at which the region is located with respect to the orientation of the parent.

Figure 5

Effect of the "sticky" attribute. This figure shows the effect of the "sticky" attribute on the placement of residues. If a position is "sticky", all the subsequent children are placed in position 0: in this way it is possible to force sub-trees in regions with a change in orientation to be drawn as a sequences (a), thus avoiding the creation of spiralling series of residues (b). The structure drawn in this figure does not represent an actual glycan.

Figure 6

Phases of the rendering algorithm, right-to-left orientation. The algorithm for computing residue bounding boxes is the core component of the rendering process. The bounding box of each residue is computed to optimize the spatial occupation of the displayed structure and to prevent collisions between nodes and edges. The algorithm recursively navigates through the tree structure performing the following phases at each step: a) compute the bounding boxes of the current residue and its sub-trees by grouping for region, b) align border regions with the current residue and place them on bottom and/or top, c) align left regions (-45, 0, 45) and place them on the left of the current residue without clashing with the border regions, d) align regions (-90, 90) with current residue and place them on bottom and/or top of the residue not clashing with the other regions, e) display results. The structure drawn in this figure does not represent an actual glycan or part of it.

Figure 7

Alignment and collision solving. Example of alignment between two sub-trees: a) align the roots of the two sub-trees on the right of their bounding boxes; b) compute the size of the shift needed to move one sub-tree on the bottom of the other maintaining a minimum distance between all their nodes; c) translate all the residues of the sub-tree according to the computed shift value.

Figure 8

Alignment of region -45, 0 and 45 with current residue, right-to-left orientation. a) If region 0 is not empty, region -45 and region 45 are aligned at right with region 0 and placed on the bottom and the top of it respectively. After that, the centre of region 0 is aligned with the centre of the current residue and the three regions are placed on the left of current residue. b) If region 0 is empty and only one region between -45 and 45 contains residues, the non-empty region is placed on the corner of the current residue, on top-left or bottom-left accordingly. c) If region 0 is empty, the regions -45 and 45 are placed on the corners of the current residue (top-left or bottom-left accordingly).

Figure 9

EUROCarbDB page of the "GlycanBuilder" applet. The "GlycanBuilder" Java applet is a visual editor for glycan structures that can be integrated in standard web pages. The applet is part of the forthcoming web interface of the EUROCarbDB database and is currently available for testing from the EUROCarbDB homepage [23].

Figure 10

"GlycoWorkbench". The "GlycoWorkbench" is an integrated suite of software tools designed for assisting the annotation of glycan fragment mass spectra during glycan sequencing. The "GlycoWorkbench" uses the "GlycanBuilder" as an internal structure editor and as a component to display structures and fragments in its various views. The application can be freely downloaded and installed from the EUROCarbDB homepage [22].