Brown University
Center for Statistical Sciences
2007 Lectureship Series
Biostatistics in the Post-genomic Era:
Statistical Methodology & Applications in Bioinformatics
Department of Statistics, University of California, Berkeley | |
Refreshments beginning at 3:15 p.m. |
Abstract: To understand gene interactions and then a more complete portrayal of gene dynamics, researchers often put together the gene expression data from multiple-experimental conditions to identify functionally related genes. In this talk, I will discuss some statistical issues concerning such studies, and introduce several non-standard methods we have developed for finding functionally related genes. In particular, I will introduce a model for capturing gene associations across multiple dependent experimental conditions. Our methods have been found to be advantageous in applications.
Sponsored by: The Charles K. Colver Lectureship and Publication Fund
Brown University
Joint Materials/Solid Mechanics Seminar Series
Mechanical Engineering Department Stanford University, CA 94305-4040 | |
Abstract: Comparing theoretical models and experimental measurements on the deformation of crystals at the micro and nano-scale is expected to bring new understanding on the fundamental nature of plasticity. Recently, there is considerable interest in the compression experiments of single crystal pillars with sub-micron diameters, because it shows a pronounced size effect in the absence of an imposed strain gradient. Several competing models have been proposed to explain this observation. Dislocation Dynamics (DD) simulations have the promise to solve this controversy because it is now possible to model all the dislocations in the micro-pillar using large scale parallel computing. We have developed an efficient way to compute the image stress due to the free surface of the micro-pillar, based on analytic solutions and Fast Fourier Transforms (FFT). DD simulation results for the dislocation behavior in FCC and BCC pillars show interesting similarity with recent experimental observations.
Note: There will be a dinner for Prof. Cai after the seminar. Please contact Ms. Pat Capece at x1501 if you wish to attend.
Brown University
Center for Computational Molecular Biology
Distinguished Technology
Lecture Series
Vice President of Informatics Affymetrix Inc. | |
Abstract: Whole genome association studies have recently become practical, both through the invention of new cost-effective genotyping technologies, and through the recent understanding of linkage disequilibrium patterns in various human populations. For example, Affymetrix GeneChip Microarrays typing 500,000 Single Nucleotide Polymorphisms have now been used to characterize loci involved in diabetes, autism, individualized drug responses, and other phenotypes of significant medical interest. Through these studies, new software tools and best practices are emerging to help with the management, quality control and analysis of these large genetic data sets. Of course, these tools become increasingly critical as even newer GeneChip technologies emerge, accessing almost a million SNPs and also providing genome-wide analysis of Copy Number Variants. We will review these new laboratory technologies and some of the computational implications of the data they produce, with a eye to the current best practices now evolving in the field. With this understanding, new research opportunities in methods and software development have become more clearly defined.
Brown University
Graduate School
Dissertation Defense Information
Applied Mathematics Graduate Student Pizza Seminar
Abstract: My idea is to give a one-hour overview of my research, focusing more on the general principles and motivations behind it rather than on mathematical detail. I will first introduce the basics of landmark-based Shape Manifolds: energy functionals, Riemannian geometry formulation, geodesic curves. Numerical results will also be shown. The second part of the talk will focus on the importance of curvature in various aspects of the analysis of shape manifolds.
PDE Seminar
Abstract: In the first part, I will discuss the general setting. Hyperbolic- parabolic systems have spatially homogeneous equilibria. When the dissipation is weak, one can derive weakly nonlinear-dissipative approximations that govern perturbations of these equilibria. These approximations are quadratically nonlinear. When the original system has an entropy, the approximation is formally dissipative in a natural Hilbert space. We show that under a mild structural hypothesis, this approximation has global weak solutions for all initial data in that Hilbert space. This theory applies to the compressible Navier-Stokes system. The resulting approximate system is an incompressible Navier-Stokes system coupled to equations that govern the acoustic modes. The solution of this approximate system is unique if the incompressible modes are uniquely determined.
In the second part, I will use the relative entropy method to prove the validity of weakly compressible Stokes approximation in the setting of DiPerna-Lions global solutions to the Boltzmann equation.
Department of Mathematics Colloquium
<--- 2007 Index