Interpreting Results

Return to Grafting documentation | Tutorials | Instructions

structures.zip contains a 3D structure for each grafted glycan. You can visualize these structures using any modeling software, including UCSF-chimera, VMD or Pymol.

If you didn’t upload a list of binders, stats.txt will just list predicted binders and non-binders respectively. If you did include a list of binders, see an example stats.txt below:

ID | Exp | CCG | Match? | Spr | Sequence
163|  N  |  N  |  Good  |  Y  | DGalpb1-4DGlcpNAcb1-3DGalpb1-4DGlcpNAcb1-3DGalpb1-4DGlcpNAcb1-OH
Correctly predicted 1 non-binders out of 1 non-binders
Correctly predicted 0 binders out of 0 experimental binders
Good Spacer effects on Exp non-binders 1
Bad Spacer Effects on Exp binders 0
Experimental binders which were not grafted:
190, 191, 192, 211, 212, 213, 216, 217, 339, 347, 349, 350, 352, 485, 50, 51, 582,
Branch-induced conformation change in 0 glycans

ID specifies the ID number used on the glycan array

Exp details ‘Y’ if you specified this glycan as a binder, and ‘N’ if it was not listed in the uploaded experimental binders file.

CCG details ‘Y’ if grafting predicts this glycan to bind, and ‘N’ if it is not predicted to bind.

Match? details ‘Good’ if the CCG and Exp columns agree, and ‘Bad’ if they do not.

Spr details ‘Y’ if grafting the spacer changed the prediction. i.e. the glycan may fit, but the spacer does not, and thus grafting predicts the glycan not to bind.

Sequence details the “sequence” or “structure” of the glycan on the array.


Example outcomes and interpretation
Exp | CCG | Match? | Spr |
 N  |  N  |  Good  |  Y  | Good as both predict non-binder. Spacer effect.
 N  |  N  |  Good  |  N  | Good as both predict non-binder. No spacer effect.
 Y  |  Y  |  Good  |  Y  | Good as both predict binder. Spacer effect.
 Y  |  Y  |  Good  |  N  | Good as both predict binder. No spacer effect.
 Y  |  N  |  Bad   |  Y  | Bad as actually a binder. Spacer effect.
 Y  |  N  |  Bad   |  N  | Bad as actually a binder. No spacer effect.
 N  |  Y  |  Bad   |  Y  | Bad as actually a non-binder. Spacer effect.
 N  |  Y  |  Bad   |  N  | Bad as actually a non-binder. No spacer effect.

Below these column headings you should see the results for each library glycan that contained the fragment you uploaded in their structure.

At the bottom stats.txt you will see:

  1. The overall agreement between the supplied binder list and grafting predictions. Important: achieving a high level of agreement (80-100%) for BOTH binders and non-binders is required for confidence in the results.
  2. Any spacer effects are also detailed, but prepare for some mental acrobatics: “Good” spacer effects occur when an experimental non-binder would be predicted to bind based on grafting the glycan alone, but the grafted spacer/surface predicted it not to bind. This is a “good” result for grafting spacers. A “bad” spacer effect is when an experimental binder would be predicted to bind by grafting, but the spacer/surface is predicted to to prevent the glycan from binding. Thus grafting the spacer reduced the level of agreement with experiment.
  3. If an experimental binder was listed in the input file, but did not contain the fragment you uploaded in its structure, it will be listed as not grafted. Only library glycans that contain the uploaded carbohydrate fragment in their sequence can be grafted!
  4. Branch-induced conformation changes arise if the full glycan structure has a significantly different shape in the carbohydrate fragment portion relative to the uploaded structure. We predict these as non-binders, as a change in the shape of the bound fragment probably inhibits recognition.

 

 

Print Friendly, PDF & Email