Figure 1.
Overview of the computational framework for glycosylation network construction.
Green solid center box presents the core functionality provided in GNAT. Surrounding this are four blue dashed boxes that highlight network construction features that are enabled by the definition of enzymes in machine-readable format.
Figure 2.
Glycosyltransferases (GTEnz) and glycosidases (GHEnz) are subclasses of the parent Enz class. A. They contain fields that are present in Enz (red) and also additional obligatory fields that describe the glycosylation reaction (green). resAtt2FG and linkAtt2FG are optional for GHEnz. In addition, the sub-class contains properties that describe additional optional details regarding enzyme specificity (blue). B. Description of the resfuncgroup, linkFG, resAtt2FG and linkAtt2FG using β1,2-N-acetylglucosaminyltransferase (GnT II) as an example. C. Examples of optional fields that constrain the substrate specificity of GnTII. D. Screenshot of a portion of the Enzyme Viewer window showing GnT II properties. Substrate specificity data are presented in LINUCS format. Clicking the ‘Plot’ button on right opens a window to display the glycan structure. Color code used to describe glycans follow the Consortium of Functional Glycomics nomenclature (www.functionalglycomics.org).
Figure 3.
Cell (depicted as dashed blue box) contains a pool of enzymes and related enzyme properties. It is possible to infer in silico: i) the intermediate reactions and products if a set of substrates is provided (‘forward network inference’); ii) the starting substrate and intermediate reactions if a set of products is provided (‘reverse network inference’); or iii) the intermediates formed given a set of starting and ending glycans (‘connection network inference’).
Figure 4.
O-linked glycosylation network construction and analysis.
A. List of glycans determined to be present in the cell based on biochemical experiments. B. Enzyme pool in cell. C. ‘Connection reaction inference’ was used to generate the ‘master pathway’ that contains all possible reactions and connections in the system. Figure generated using the glycanPathViewer function of GNAT. D. Path finding function determines the reaction pathway between the core-2 trisaccharide (glycan 0) and glycans containing the sLeX epitope (glycans 5 and 9). E. Histogram plot quantifying the number of subset pathways generated when 1 to 10 O-glycans were deleted from the master pathway in panel C.
Figure 5.
N-linked glycosylation pathway reconstruction.
A. Enzyme rules are defined for a pool of six enzymes: GnT II, III, IV, V, ManII and GalT. ‘-’ indicates blank fields in the GHEnz/GTEnz class definitions. B. N-linked glycosylation pathways generated using the ‘forward network inference’ function, when GlcNAcMan5GlcNAc2 was the starting substrate and five enzymes defined in panel A (GnT II-V and Man II) were used. C. Network generated using all six enzymes in panel A.
Figure 6.
Reaction network generated from glycomics experimental data.
A. The readMS and msprocess functions of GNAT were used to smooth raw experimental data. B. Glycans annotated in the MS spectra available from the Consortium of Functional Glycomics web site. C. ‘Connection network inference’ applied for network construction using 43 annotated input glycans and 9 enzymes (GnT I-V, ManII, FucT, SiaT, and GalT). Constructed network has 360 reactions and 188 species. D. Expanded view of a portion of the network. E. pathfinding function used to connect the starting GlcNAcMan3GlcNAc2 structure and the tri-bisecting antennary glycan with m/z = 2937.9.