The ¹³C NMR spectrum assignment table prediction feature is proposed for the simulation of the ¹³C chemical shifts (in water solution) for the given fragment of the glycan structure. Together with chemical shift assignment table it outputs the substitution effects and the spectra of free residues for reference. The simulated assignment table and other information can be copied and pasted. The feature is available from the search results page and from the structure wizard. The structure is virtually de-O-acetylated before processing.

Prediction accuracy

The overall consistency of prediction is the average of all residues prediction consistencies. It utilizes the range from 0 to 4, the higher the more accurate prediction is. Spectra predicted with total consistency above 3.0 may be considered quite trustworthy. The consistency of prediction of each residue subspectrum may be from 0 to 4 as well, depending on the following:

• consistency=4 is the best variant and implies one of two:
  • Exact data for this residue (substituted exactly as in the structure) has been found in the primary NMR spectroscopic database.
  • The residue is monosubstituted and its subspectrum has been simulated from the spectrum of the free residue and the effect of substitution specific for the exact structural enclosement of this residue. The effect has been found in the substitution effect database.
• consistency=3 covers the following cases:
  • The residue is monosubstituted and its subspectrum has been simulated from the spectrum of the free residue and the effect of substitution specific for the structural enclosement of this residue which is similar to that found in structure. The structural enclosement was varied until the exact effect has been found in the substitution effect database.
  • The residue is bisubstituted at distant positions and its subspectrum has been simulated from the spectrum of the free residue and the effects of monosubstitution specific for the exact structural enclosement of this residue. Both effects has been found in the substitution effect database and applied additively.
• consistency=2 covers the following cases:
  • The residue is bisubstituted at distant positions and its subspectrum has been simulated from the spectrum of the free residue and the effects of monosubstitution specific for the structural enclosement of this residue which is similar to that found in structure. The structural enclosement was varied until both exact effects have been found in the substitution effect database and applied additively.
  • The residue is bisubstituted at neighbouring positions or multisubstituted (tri- or more) and its subspectrum has been simulated from the spectrum of the free residue and the effects of monosubstitution specific for the exact or similar structural enclosement of this residue (i.e. with structure variation or without). Both effects has been found in the substitution effect database and applied additively.
• consistency=1 means that at least one of the substitution effects on this residue could not been found in the substitution effect database, with or without structural enclosement variantion. This effect has been simulated as +6.0 (glycosidic) or +3.0 (amidic) on alpha-carbons, -1.0 or none on beta-carbons, +6.0 on the linking atom of the substituent.
• consistency=0 is the worst variant. The prediction engine could not find the primary NMR data for this residue. The residue is excluded from further consideration except that effects on the linking atom of its substituents are counted as +6.0. The resulting spectrum will lack signals of this residue. To obtain the better prediction accuracy, try removing O-acetyl groups or other monovalent substituents from this residue.

NMR databases and algorythm

The primary spectroscopic database and the substitution effect database contain averaged literature data on chemical shifts, glycosylation and phosphorylation effects. The approximate coverage is 80 residues, 2500 dimers and trimers, 150 effects; data are averaged for D₂O solutions at 318K. The following structural peculiarities are taken into account when searching for a particular chemical shifts or substitution effects:

• which residue is substitited
• its anomeric configuration, if exists
• the substitution position
• the type of substituent: pyranose, furanose, alditol, phosphate, non-carbohydrate
• anomeric configuration, if exists
• additional groups attached to C1 (none, carboxyl group, C-chain)
• the type of group at substituent's C2 (-OH/-NH2/deoxy)
• the orientation of proton at substituent's C2 (axial/equatorial/unoriented or both)
• the combination of absolute configurations of the substituent and the residue itself (same or different)

The NMR predictor iterates through all residues in the structure and searches the primary spectroscopic database for chemical shifts characteristic for this residue in given structural enclosement. If these data are not found, the subspectrum of the residue is calculated from the spectrum of the free residue and substitution effects. Please note, that primary NMR spectroscopic database does not contain data on non-N-acetylated residues, thus you have to specify aminosugars as N-acetyled to obtain the better accuracy.

If the desired effect is missing from the database, the type and orientation of the substituent C2 position are varied until the effect is found. If the effect is found for none of the variants the residue being predicted is temporarily replaced with a common residue with the same basic configuration (e.g. Gal instead of FucNAc). If still no effect is found, it is simulated as +6.0 on alpha-carbon, -1.0 on beta-carbons, +6.0 on C1 of the substituent (for O-linked residues) or as -3.0 on alpha-carbon, +1.0 on beta-carbons, +3.0 on C1 of the substituent (for N-linked residues).

If a residue subspectrum is calculated using substitution effects, chemical shifts of C2 and C5 of pyranoses that are attached to anything by their C1 are modified the following way: C2 -1.4 for b-Gal, C2 -1.4 for other beta-sugars, C2 -0.5 for alpha-sugars with equatorial H2, C5 +0.5 for all alpha-sugars.

The glycosylation effects for three widespread sugar configurations (glc, gal, man) are represented most completely, usually making the effect prediction for these basetypes more accurate.

Acknowledgement and reference

The ¹³C NMR prediction feature was made on the base of the spectra simulation module from BIOPSEL software [1,2]. Within BCSDB project it was adapted to oligomeric structures, improved to treat keto-sugars and other 'special cases' and got web-interface.

1. F.V. Toukach, A.S. Shashkov "Computer-assisted structural analysis of regular glycopolymers on the basis of ¹³C NMR data" (Carbohydrate Research, 2001, v.335(2), pp.101-114)
2. F.V. Toukach "Computer-assisted structural analysis of glycopolymers" (Proceedings of Eurocarb-12, France, Grenoble, 2003, p. PA-004)