Integration Of Data

A key objective in toxicogenomics is to integrate data from different studies and analytical platforms to produce a richer and biologically more refined understanding of the toxicological response of a cell, organ, or organism. For example, one would like to describe the interplay between protein function and gene expression, or between the activity of certain metabolizing enzymes and the excretion into serum or urine of populations of small metabolites. The integration of data from different domains such as proteomics and transcrip-tomics [54,97,138], or transcriptomics and metabonomics [46], has been reported. In these experiments, tissue samples derived from the same individual animals or from comparably treated animals were analyzed in parallel using different technologies. However, the data from different studies were integrated only after a short list of differentially responsive transcripts or protein spots had been obtained.

The experience gained from integrating global proteomics or metabonomics data, such as spot intensities from 2D gels or metabonomics fingerprint data from NMR, tells us that cluster or principle component analysis can be performed to obtain global signatures of molecular expression in much the same way as in transcriptomics analyses. If biological samples segregate into unique clusters that show similar expression characteristics, with additional effort the novel proteins or metabolites that are expressed in these samples can be discerned. Steps can then be taken to evaluate these proteins or metabolites as potential biomarkers and thus to determine the underlying toxicological response.

Although software is plentiful for managing expression profiling data at the laboratory level, there is a great need for public databases that combine profile data with associated biological, chemical, and toxicological endpoints [1]. Comparisons of gene, protein, and metabolite data in public databases can be a valuable tool and assist in global understanding of how biological systems function and respond to environmental stressors [65,139]. As these repositories are developed, experiments will be deposited from disparate sources, with different experimental designs and yet targeting the same toxicity endpoint or a similar class of toxicant. In these cases it will be important that the databases integrate data from related studies before performing data mining. To maximize the value of deposited datasets, the repositories must also be able to integrate data from different technological domains (see Appendixes 5.1, 5.2, and 5.3). Members of regulatory bodies are working with scientists from industrial, academic, and governmental laboratories participating in the ILSI Genomics Committee and Clinical Data Interchange Standards Consortium/ Standards for Exchange of Nonclinical Data (CDISC/SEND) Consortia to develop standards for the exchange, analysis, and interpretation of transcrip-tomics data.

A proposal has been made to extend toxicogenomics and combine it with computational approaches such as physiologically based pharmacokinetic (PB/PK) and pharmacodynamic modeling [26]. PB/PK modeling can be used to derive quantitative estimates of the dose of the test agent or its metabolites that are present in the target tissue at any time after treatment, thereby allowing molecular expression profiles to be anchored to internal dose, as well as to the time of exposure and to the toxicant-induced phenotype. Relationships among gene, protein, and metabolite expression can then be described both as a function of the applied dose of an agent and the ensuing kinetic and dynamic dose-response behaviors that occur in various tissue compartments. Such models also must take into account the fact that the transcriptome, pro-teome, and metabolome are themselves dynamic systems, and are therefore subject to significant environmental influences, such as time of day and diet [140-142].

Despite the numerous successes of toxicogenomics in the context of toxicology, a poorly addressed but confounding issue pertinent to drug safety and human risk assessment is the impact of the individual genetic background on the response of the individual animal or human patient. The PharmGKB pharmacogenetics knowledgebase [143] cataloges the different human genetic backgrounds by their susceptibility to drug therapy. In addition the NIEHS Environmental Genome Project (EGP [24]) is identifying SNPs in genes that are important in environmental disease, detoxification, and repair. Linkages of toxicogenomics knowledgebases with those containing information about SNPs and human susceptibility will gradually lead to a more complete picture of the relevance of the responses and genotypes of surrogate animal species to human risk assessment.

0 0

Post a comment