Here are three major results obtained so far in this research project:? ? 1. For over two decades the National Toxicology Program (NTP) has been interested in developing a formal statistical procedure for analyzing data from their 2-year rodent cancer bioassay that incorporates their historical control data base. Several statistical procedures have been proposed over the past two decades but the Technical Reports Subcommittee of the NTP Board of Scientific counselors has not endorsed any of the existing methods and recommended that a new procedure be developed for this important problem. In response to this requirement, we developed a new statistical procedure for this important long standing problem. The resulting paper was accepted for publication by the Journal of American Statistical Association. I am also in the process of developing user friendly software based on this methodology. We are currently evaluating the performance of this procedure so that it can be used for analyzing future NTP bioassays. ? ? 2. During the past few years several statistical procedures have been developed to analyze time-course, dose-response gene expression microarray data. One such method is ORIOGEN which was developed by me and my colleagues at NIEHS. Most methods (including ORIOGEN) assume that the variability in the gene expression of a gene over time is constant. This assumption need not be true in general. In collaboration with Professor Susan Simmons, UNC-Wilmington, I introduced a modification to ORIOGEN which relaxes the above assumption and allows for heterogeneity of gene expression over time or dose. ? ? 3. Often researchers at NIEHS (and at other places) collect data from dose-response, time-course experiments where the outcome variable is ordinal. In such cases the existing methods of analysis are not satisfactory, as they can be severely under-powered. Motivated by a data set obtained received in our consulting service, we developed a new general methodology which is applicable to a very broad collection of data sets including ordinal data and two-dimensional gene expression microarray data.
Showing the most recent 10 out of 36 publications
