Quantitative Analysis of Lectin Microarray Data Using Simplified Binding Models Derived from Glycan Array Data
Location
Loosemore Auditorium
Description
PURPOSE: The analysis of lectin array data is typically performed by manually and qualitatively assessing binding to each lectin individually. Such analysis does not consider the complexities of each lectin’s specificity nor the integrated information from lectins with overlapping specificities. We previously developed software algorithms to provide this functionality, but the algorithms were limited by oversimplified assumptions of the relationship between glycan binding and signal as well as insufficient calibration of glycan binding. In this work, we introduce a method for solving glycan motif abundance using binding kinetics through a simplified competitive binding model for multiple ligands with binding properties derived from glycan arrays. In this way, we accurately model the relationship between binding and signal while retaining a computationally feasible model. PROCEDURES: To enable this analysis, we first developed a system for calibration of lectin arrays which places binding on a common scale. In addition, we developed a system of constraints that encode expert biological knowledge into the fitting process. We tested this approach on the analysis of the well-characterized glycoproteins fetuin and asialofetuin by 37 lectins. OUTCOME: The quantifications of glycan motif abundance correlated with orthogonal, historical data with a coefficient of 0.70 for Fetuin and 0.72 for Asialofetuin, demonstrating a substantial improvement over previous method. IMPACT: Thus, the incorporation of competitive binding models, calibration, and expert knowledge may dramatically improve the automated analysis and interpretation of lectin array data, ultimately increasing the value of lectin array technology.
Quantitative Analysis of Lectin Microarray Data Using Simplified Binding Models Derived from Glycan Array Data
Loosemore Auditorium
PURPOSE: The analysis of lectin array data is typically performed by manually and qualitatively assessing binding to each lectin individually. Such analysis does not consider the complexities of each lectin’s specificity nor the integrated information from lectins with overlapping specificities. We previously developed software algorithms to provide this functionality, but the algorithms were limited by oversimplified assumptions of the relationship between glycan binding and signal as well as insufficient calibration of glycan binding. In this work, we introduce a method for solving glycan motif abundance using binding kinetics through a simplified competitive binding model for multiple ligands with binding properties derived from glycan arrays. In this way, we accurately model the relationship between binding and signal while retaining a computationally feasible model. PROCEDURES: To enable this analysis, we first developed a system for calibration of lectin arrays which places binding on a common scale. In addition, we developed a system of constraints that encode expert biological knowledge into the fitting process. We tested this approach on the analysis of the well-characterized glycoproteins fetuin and asialofetuin by 37 lectins. OUTCOME: The quantifications of glycan motif abundance correlated with orthogonal, historical data with a coefficient of 0.70 for Fetuin and 0.72 for Asialofetuin, demonstrating a substantial improvement over previous method. IMPACT: Thus, the incorporation of competitive binding models, calibration, and expert knowledge may dramatically improve the automated analysis and interpretation of lectin array data, ultimately increasing the value of lectin array technology.