• 10ug/mL
• 1ug/mL
• 0.1ug/mL
• Kd

Figure 3. Boxplots of glycans grouped by motif. Individual points are glycans. Mouseover to see glycan structures.

Motif ID	Nearest Common Name (Accuracy%**)	Motif Graphic Structure	Relative Binding	Number of Glycans	P-Value***
A2*	Terminal Mannose Terminal Type 2 LacNAc (97%)		1.00	2	0.000
A0*	Terminal Mannose Terminal N-Actyl Hexosamine (97%)		0.88	3	0.000
A3	Terminal Mannose a3 Sialyl Type 2 LacNAc (96%)		0.43	2	0.009
A1	Terminal Mannose a3 Sialyl Type 2 LacNAc (96%)		0.41	2	0.016
B0*			0.63	4	0.000
B1			0.10	66	0.192
0	Non-Binders (100%)		0.00	19	NA

Figure 3. Boxplots of glycans grouped by motif. Individual points are glycans. Mouseover to see glycan structures.

Motif ID	Nearest Common Name (Accuracy%**)	Motif Graphic Structure	Relative Binding	Number of Glycans	P-Value***
A2*	Terminal Mannose Terminal Type 2 LacNAc (97%)		1.00	2	0.000
A0*	Terminal Mannose Terminal N-Actyl Hexosamine (97%)		0.99	3	0.000
A3	Terminal Mannose a3 Sialyl Type 2 LacNAc (96%)		0.23	2	0.232
A1	Terminal Mannose a3 Sialyl Type 2 LacNAc (96%)		0.16	2	0.645
B0*			0.52	4	0.000
B1			0.06	66	0.567
0	Non-Binders (100%)		0.00	19	NA

Figure 3. Boxplots of glycans grouped by motif. Individual points are glycans. Mouseover to see glycan structures.

Motif ID	Nearest Common Name (Accuracy%**)	Motif Graphic Structure	Relative Binding	Number of Glycans	P-Value***
A0*	Terminal Mannose Terminal N-Actyl Hexosamine (97%)		1.00	3	0.000
A2*	Terminal Mannose Terminal Type 2 LacNAc (97%)		0.46	2	0.000
A3	Terminal Mannose a3 Sialyl Type 2 LacNAc (96%)		0.06	2	0.959
A1	Terminal Mannose a3 Sialyl Type 2 LacNAc (96%)		0.02	2	1.000
B0*			0.26	4	0.000
B1			0.03	66	0.915
0	Non-Binders (100%)		0.00	19	NA

Figure 3. Boxplots of glycans grouped by motif. Individual points are glycans. Mouseover to see glycan structures.

Motif ID	Nearest Common Name (Accuracy%**)	Motif Graphic Structure	Kd	Number of Glycans	P-Value***
A0*	Terminal Mannose Terminal N-Actyl Hexosamine (97%)		1.10244	3	0.396
A2*	Terminal Mannose Terminal Type 2 LacNAc (97%)		1.14305	2	0.482
A1	Terminal Mannose a3 Sialyl Type 2 LacNAc (96%)		2.54177	2	0.503
A3	Terminal Mannose a3 Sialyl Type 2 LacNAc (96%)		2.86257	2	0.508
B0*			2.39995	4	0.367
B1			55.689	38	1.000
0	Non-Binders (100%)		62.6552	2	NA

*Motif indicates the remaining glycans not matched by motifs which are a subset. Motif definition needs to be taken in the context of the model.

**Accuracy describes the consistency between common-name definition of the motif and the formal, text-based definition of the motif, in terms of percent agreement in the glycans containing the two motifs. Common Name label definitions given here.

***P-Value refers to difference from Non-Binders with multiple testing correction (Dunnet’s Test)

Motifs with a red motif ID fail to show a logistic response to protein concentration in the range of concentrations analyzed. These motifs may be nonbinding motifs (motifs which define nonbinding exceptions) or simply fail to fit. Nonbinding motifs are determined based on concentration dependent response when available or the average binding of non-motif glycans otherwise.

Figure 4. Glycan intensity and motif distribution plot. The top half of the plot presents the observed glycan binding intensity of various glycans used in the array over their rank binding intensity; only the top glycans are shown. The second plot indicates the position of glycans containing the various motifs in the top plot with a yellow tick.

*Motif indicates the remaining glycans not matched by motifs which are a subset. Motif definition needs to be taken in the context of the model.

Figure 5. Treemap of glycan binding grouped by motif and family structure. The model structure can be represented as nested boxes where box size is proportional to the number of glycans with the motif and color changes with change in average relative binding of glycans with the motif. Only three layers of data splitting are included here, though further splitting may be possible.

*Motif matches the remaining glycans not matched by earlier motifs in the model.

Figure 6. Tree representation of the regression tree model trained on array data. Data flows through the tree (top-down) and is split by the various motifs. The motif used the split the data at each point has the id “family+split number” except when further split. In the case of futher splits the id of the motif used to split the data is denoted with an asterisk.

Figure 7. Logistic curves fit to average motif binding and glycan binding. Curves are fit with as many parameters as possible given the data. All curves are based on the logistic curve with a fixed intercept of 0. Mouse-over to view curve info and highlight glycans/motifs on the other plot.

Figure 8. Scatterplots of logistic curve parameters for glycans (points) and motifs (text).