We developed a quantitative, statistically sound methodology for the analysis of suspected gene regulatory networks using gene expression data sets. The method is based on Bayesian networks and provides a means to directly quantify gene-expression networks and test hypotheses regarding the linkages between genes in this network. The method is currently being applied to a number of data sets including data on early brain development and dioxin toxicity.? ? Data on the molecular scale obtained under different sampling conditions are becoming increasingly available from systems like DNA microarrays. These data can be related to each other in a number of ways such as through their gene products (proteins, enzymes) being involved in the same metabolic andor signaling pathways, through their common linkage to certain diseases, or simply because their parent genes have similar chromosomal locations. In the interpretation of these data, a key question is how to intelligently propose ideas and experiments for further studies based not only on the changing molecular measures, but on their biologically-relevant pathways. We use two biologically-inspired hypotheses that would help identify pathways that would be more relevant to a certain disease condition or the presence of a toxin. First, pathways where the gene products of the perturbed genes are close together in the inherent flow of materialinformation of these pathways are worth considering for further analysis. And second there are certain gene products at the beginning of the biochemical pathway, like the receptorsgrowth factors that are more important than others in being responsible for turning on a pathway. Using these hypotheses, we are developing a novel pathwaygene set enrichment algorithm. We demonstrate its efficacy using different sets of simulated data and also with microarray data obtained from patients with Type 2 diabetes, when compared with the methods currently used.? ? We are also developing methodology to test the hypothesis of being able to predict genetic response under certain conditions using a database of gene expression responses to carefully chosen genetic perturbations. ? ? In PBPK modeling, the effects of model mis-specification with respect to compound transport into tissues has been investigated with focus on experimental design, coverage by joint confidence regions of the true values of estimated parameters, and the alpha levels of likelihood ratio tests, AIC and BIC tests for model selection.? ? We have a focused effort on environmental components that may effect brain development. Our initial project is a collaboration between our lab and Fthe National Institute of Medical Research in London, and involves the analysis of genomic data during telencephalon neurogenesis in the mouse. Several experimental and computational analyses were added to the initial submission of this research, based on reviewers suggestions. Building from this project of genetic manipulation during neurogenesis, we have undertaken a project in which AhR transgenic mice were exposed to TCDD in utero, and dorsal and ventral telencephalon were dissected via laser capture. Currently, RNA is being isolated for microarray hybridization. ? ? We have also initiated a collaboration with NIA on expanding the utility a database of polymorphisms associated with particular phenotypes in human populations. We are using this dataset, along with gene-environmental factor data from the chemical toxicogenomics database, to discover relationships between genes, pathways, diseases, and environmental factors. ? ? We have designed and analyzed a series of experiments running beads through the COPAS Biosorter. The analysis shows that measurements on beads was reproducible with consistent machine settings as well as the effect of the machine settings on two classic measurements, EXT and TOF. A pronounced batch effect on bead measuremennts was also uncovered. We also applied a previously developed Markov type population model for growth to 2 experiments exposing C. elegans to chlorpyrifos. The estimated growth rates were used to deduce the nature of the chlorpyrifos effect on growth. Later we extended the model to include death of worms and the curling of worms as it affected TOF measurements.? ? An algorithm was developed for estimating absolute quantities of initial amplicon in PCR amplifications. This method is an improvement over other analytical approaches because it allows the absolute estimate to be made using a single amplification run. ? ? A rigorous analysis is being conducted comparing the previously developed Tao-Gen algorithm to current alternative Bayesian Network methods with respect to alternative prior distributions. The applications area is primarily uncovering geneprotein signal transduction networks with missing nodes. ? ? We are developing a mathematical and pharmacological framework to describe the absorption, distribution and metabolism of nanoscale materials in mammalian systems. This work is in collaboration with colleagues at the National Toxicology Program, the National Center for Toxicological Research and North Carolina State University. Early indications are that size does matter in dispersal to the brain and in half-life in the body.? ? A physiologically based pharmacokinetic model was developed to represent the absorption, distribution, metabolism, and excretion of anthraquinone in rats. The model results were used in a two-stage model of carcinogenesis applied to results from the chronic feed study of anthraquinone. ? ? Cell viability high throughput screening (HTS) data for 13 cell lines were analyzed to screen 1353 chemicals for dose-response effects. The analysis revealed systematic effects of plate location on the measured cell viability. A dose-response analysis of the data was performed with an algorithm including corrections for the plate location effects. Various statistical tests for identifying significant concentration-response relationships and for addressing replicability were developed.
Showing the most recent 10 out of 57 publications
