The Biostatistical Research Core (BRC) will continue to serve the six projects through applications of classical, modern, and newly-emerging quantitative methods for experimental design and analysis of empirical findings. In addition, the BRC will work with the Computational and Animal Behavioral Cores to develop new computational and biostatistical technologies for the projects. The BRC will also serve to integrate the efforts of all three cores. The major continuing challenge of the BRC is to push biostatistical interactions with the projects beyond mere confirmation of their expectations towards the discovery of previously undetected signals hidden in noisy data. With the Benes project and the Animal Behavioral Core, the BRC will apply mixed-effects Poisson analyses of variance models to test the hypotheses that GABA blockade and reduced NMDA receptor function in hippocampal CA sectors can result in aspects of cognitive dysfunction found in schizophrenia by assessing the significance of Treatment X Environment interactions in the novelty detection study. Computer-intense Bayesian Markov Chain Monte Carlo (MCMC) methods will also be employed. In a combined analysis, the DISH experiments will employ similar yet weighted Poisson models with silver grain NR2A densities as weights for the counts of GAD67-positive hippocampal interneurons. We will also apply our existing semi-automatic algorithms for detection of Fos bodies. With the Lisman project and the Computational Core, we will apply MCMC technology to revise prior connectivitymorphology probability mappings based on existing literature by new data likelihoods obtained from studies in the Benes project to yield updated posterior probability mappings. The Greene project and the Animal Behavioral Core will benefit from our longitudinal mixed-effects logistic regression models applied to their odor familiarity and recognition study. The BRC will serve as consultant to the Coyle, Yurgelun-Todd and Goff projects on an as needed basis. Further biostatistical applications include tests to distinguish between local spatial heterogeneity (different neuronal densities in different regions) and spatial dependence (correlations between neuronal locations) and Poisson random field methods to detect features of 3D brain cell assemblies at multiple levels of spatial resolution.
Showing the most recent 10 out of 128 publications
