Informatics support for human membrane transporter pharmacogenomics studies

My research has been trying to solve some fundamental problems in two major biomedical informatics subjects. One is the knowledge domain of pharmacogenomics, applied specifically to human membrane transporters. This is a relatively new area that lacks informatics studies and available methodologies. The multidisciplinary characteristics of this area have barriers between disciplines for informatics support. The huge amount and the heterogeneous nature of the transporter data demand serious effort of organization to support drug targeting and microarray studies.; The other domain is bioinformatics itself. In this area, lack of common literacy has been recognized as one of the most important obstacles in bioinformatics development.; A series of methodologies have been developed to solve these problems. First of all, three correlations, structure-function, gene-drug, and genotype-phenotype, were found as fundamental themes for understanding pharmacogenomics. The analysis of these correlations clarifies the complexity of some problems and defines the domain boundary.; With the correlations as major targets, data modeling was attempted to lay the ground for further data analysis and system implementation. To this end, a three-step methodology has been developed to create an object-oriented (OO) model for pharmacogenomics. This methodology was designed to overcome the barriers in different disciplines and to establish a standard methodology in pharmacogenomics informatics studies.; A database powered decision support system (DSS) was developed to implement this OO model. This DSS has a three-tiered client-server architecture and is accessible from the Web to provide broad applications. The system user interface was designed with domain expertise to make its usage easier and more efficient.; Because of the heterogeneous nature of transporter data, specific methodologies for data collection and integration were developed to support pharmacogenomics studies. The data integration process addressed some special features to provide the most complete information about transporters, including classification and terminology.; The methodologies developed here also provide generic solutions for problems in the bioinformatics domain. The successful implementation of the system demonstrated that applying unified modeling language (UML) to biomedicine is a generic bioinformatics methodology. For the first time, this work recognized bioinformatics design patterns that can be reused for building more bioinformatics systems.