Analysis of pharmacokinetics, pharmacodynamics, and pharmacogenomics data sets using VizStruct, a novel multidimensional visualization technique

• Published in: Pharmaceutical research Vol. 21; no. 5; pp. 777 - 780
• Main Authors: Bhasi, Kavitha, Zhang, Li, Zhang, Aidong, Ramanathan, Murali
• Format: Journal Article
• Language: English
• Published: United States Springer Nature B.V 01-05-2004
• Subjects: Algorithms; Animals; Case studies; Database Management Systems; Databases, Factual; Datasets; Fourier transforms; Gene Expression Profiling; Genes; Humans; Kinetics; Pharmaceutical sciences; Pharmacodynamics; Pharmacogenetics - statistics & numerical data; Pharmacokinetics; Pharmacology - statistics & numerical data; Population; Software; Visualization
• ISSN: 0724-8741; 1573-904X
• DOI: 10.1023/B:PHAM.0000026427.30177.61

Data visualization techniques for the pharmaceutical sciences have not been extensively investigated. The purpose of this study was to evaluate the usefulness of VizStruct, a multidimensional visualization tool, for applications in pharmacokinetics, pharmacodynamics, and pharmacogenomics.
The VizStruct tool uses the first harmonic of the discrete Fourier transform to map multidimensional data to two dimensions for visualization. The mapping was used to visualize several published pharmacokinetic, pharmacodynamic, and pharmacogenomic data sets. The VizStruct approach was evaluated using simulated population pharmacokinetics data sets, the data from Dalen and colleagues (Clin. PharmacoL Ther. 63:444-452, 1998) on the kinetics of nortriptyline and its 10-hydroxynortriptyline metabolite in subjects with differing number of copies of the CYP2D6, and the gene expression profiling data of Bohen and colleagues (Proc. Natl. Acad. Sci. USA 100:1926-1930, 2003) on follicular lymphoma patients responsive and nonresponsive to rituximab.
The VizStruct mapping preserves the key characteristics of multidimensional data in two dimensions in a manner that facilitates visualization. The mapping is computationally efficient and can be used for cluster detection and class prediction in pharmaceutical data sets. The VizStruct visualization succinctly summarized the salient similarities and differences in the nortriptyline and 10-hydroxynortriptyline pharmacokinetic profiles in subjects with increasing number of CYP2D6 gene copies. In the simulated population pharmacokinetic data sets, it was capable of discriminating the subtle differences between pharmacokinetic profiles derived from 1- and 2-compartment models with the same area under the curve. The two-dimensional VizStruct mapping computed from a subset of 102 informative genes from the Bohen and colleagues data set effectively separated the rituximab responder, rituximab nonresponder, and control subject groups.
The VizStruct approach is a computationally efficient and effective approach for visualizing complex, multidimensional data sets. It could have many useful applications in the pharmaceutical sciences.