Brown University Center for Statistical Sciences Seminar
National Cancer Institute, Division of Cancer Epidemiology and Genetics | |
Refreshments at 3:45 pm |
Abstract: Absolute risk is the probability that an individual who is free of a given disease at an initial age , a, will develop that disease in the subsequent interval (a, t]. Absolute risk is reduced by mortality from competing risks. Models of absolute risk that depend on covariates have been used to design interventions studies, to counsel patients regarding their risks of disease, and to inform clinical decisions, such as whether or not to take tamoxifen to prevent breast cancer. Several general criteria have been used to evaluate models of absolute risk, including how well the model predicts the observed numbers of events in subsets of the population ("calibration"), and "discriminatory power", measured by the concordance statistic (e.g. Rockhill et al., J Natl Cancer Inst, 93, 358-366, 2001). We review some general criteria and develop specific loss function-based criteria for two applications, namely whether or not to screen a population to select subjects for further evaluation or treatment and whether or not to use a preventive intervention that has both beneficial and adverse effects. We find that high discriminatory power is much more crucial in the screening application than in the preventive intervention application. These examples indicate that the usefulness of a general criterion such as concordance depends on the application, and that using specific loss functions can lead to more appropriate assessments.
Sponsored by the Charles K. Colson Lectureship and Publication
Fund
Co-Sponsored by the Bruce M. Bigelow Class of 1955 Lecture Series
Stochastic Systems Seminar
Abstract: Importance sampling (which is a variance reduction technique for stochastic simulation) has become a standard tool in estimating probabilities of rare-events. Recently, in work by Chen, Diaconis, Holmes and Liu (2005), this technique was also applied to approximate counting problems (the number of binary contingency tables). Although empirical evidence suggests that importance sampling may be suitable technique for approximate counting, the theoretical support to substantiate this evidence is very limited.
We shall discuss a new technique, based on Lyapunov bounds for Markov chains, that can be used to analyze the computational complexity of state-dependent importance sampling algorithms (for rare-events and counting). As an application, we analyze the computational complexity of Chen et al (2005)'s counting algorithm in the context of large and sparse binary contingency tables. We will also illustrate this technique in the context of rare-event simulation. In particular, we shall present an algorithm (joint work with Peter Glynn) that can be proved to be efficient (in a suitable sense) for estimating the tail of the maximum of a general random walk -- which has important implications in queueing and insurance. This is the first algorithm that is known to be probably efficient for the maximum of random walks with general heavy-tailed distributions (we only require subexponentiality of the right tail and the corresponding integrated tail).
Brown University
Graduate School Dissertation Defense
Special LCDS/PDE Seminar
Abstract: In this talk we will survey the recent mathematical theories for Boltzmann equation, in particular, the Green's functions for Boltzmann equation. There are several highlights in this talk: macro-micro decomposition, positive solution of the Boltzmann shock layer, hydrodynamic limit problem, Boltzmann boundary layer, particle-wave decomposition, mixture lemma, and the Green's functions for both initial value problem and initial-boundary value problems.
The Center for Computational Molecular Biology Seminar
Refreshments at 3:45 p.m. |
Abstract: With more than 300 bacterial and archaeal genomes completely sequenced and the total sequence content of GenBank still growing exponentially, we can now gain some impression of the distribution of RM systems in the real world. Surprisingly, these RM systems, or the relics of them, are much more abundant than might have been guessed from the classical biochemical screening of strains in the laboratory. In particular, Type I systems are widely distributed in Nature and many instances of solitary specificity subunits are found. More than 100 potential Type III and Type IV systems are found and on average about 4 DNA methyltransferase genes are found per genome. Solitary M genes, in which the R gene is either missing or non-functional, are much more common than expected. However, our ability to identify M genes accurately is made difficult by the presence of conserved motifs in genes that methylate molecules other than DNA. Analyses of the many environmental samples in GenBank suggests that the rate of evolution of both M and R genes is quite high and confirms previous findings that the direct cloning of intact RM systems into E. coli is difficult with current technology. Importantly, there is little reason to think that our current collection of 256 Type II specificities is more than a small sample of the specificities present in the environment.
Dr. Richard Roberts is the Chief Scientific Officer at New England Biolabs, Ipswich, MA. He was educated in England, attending the University of Sheffield where he obtained a B.Sc. in Chemistry in 1965 and a Ph.D. in Organic Chemistry in 1968. His postdoctoral research was carried out in Professor J.L. Strominger's laboratory at Harvard, where he studied the tRNAs that are involved in the biosynthesis of bacterial cell walls. From 1972 to 1992, he worked at Cold Spring Harbor Laboratory, reaching the position of Assistant Director for Research under Dr. J.D. Watson. He began work on the newly discovered Type II restriction enzymes in 1972 and in the next few years more than 100 such enzymes were discovered and characterized in Dr. Roberts' laboratory. Dr. Roberts has been involved in studies of Adenovirus-2 and discovered split genes and mRNA splicing in 1977 for which he received the Nobel Prize in Physiology or Medicine in 1993. His laboratory sequenced the 35,937 nucleotide Adenovirus-2 genome, and wrote some of the first programs for sequence assembly and analysis. DNA methyltransferases are an area of active research interest and, in collaboration with Dr. X. Cheng, DNA base flipping was discovered in 1993. Current interests focus on the identification of restriction enzyme and methylase genes within the GenBank database and the development of rapid methods to assay their function.
Hosted by: Sorin Istrail
The Center for Computational Molecular Biology Seminar
Refreshments served at 2:45 pm |
Abstract: The recent release of the Haplotype Mapping project (Nature, Oct. 26, 2005), and the rapid reduction in genotyping costs open new directions and opportunities in the study of complex genetic disease such as cancer or Alzheimer's disease. The datasets collected for many of these studies include Single Nucleotide Polymorphisms (SNP) data, which are DNA sequence variations that occur when a single nucleotide (A,T,C,or G) in the genome is altered.
Even though technological improvements have recently reduced the genotyping costs considerably, the genotyping burden on disease association studies is still heavy. One technique that may be able to reduce this burden is the use of DNA pools. In DNA pools, the DNA samples of a group of individuals is pooled, and the resulting pool is then genotyped, resulting in a measure of the allele frequency in the pool. In this talk, I will describe new methods that use DNA pools for association studies involving unrelated individuals or mother-father-child trios. I will show how some combinations of DNA pools can reduce the genotyping burden considerably, or alternatively, can serve as "error detecting codes". I will also describe some wet lab experiments that support these results.
Hosted by: Sorin Istrail
Center for Fluid Mechanics
And
The Fluids, Thermal And Chemical Processes Group
Of
The Division Of Engineering
Seminar Series
Department of Mechanical and Aerospace Engineering and Whitaker Institute of Biomedical Engineering, University of California, San Diego, La Jolla, CA | |
Barus and Holley, Room 190 |
Abstract: Cell motility is essential for a variety of processes such as vertebrate embryonic development, tissue repair, and the metastatic spreading of some cancer cells. Cell crawling requires the integration and coordination of complex biochemical and biomechanical signals which regulate the traction forces exerted through the cytoskeleton at the local adhesion points on the substratum. Although it is generally accepted that actin and myosin are the common elements in most cell crawling movements, these proteins undergo many different transformation as the cell migrates. A complete understanding of the mechanisms of cell crawling must provide a molecular explanation for these transformations, explain how they are coordinated in time and space, and relate them to changes in the mechanical parameters such as the rheological properties of the cell and the adhesion forces to the substrate. This lecture will discuss new measurements of the spatial and temporal distribution of traction forces as well as the associated changes in the mechanical properties of the cell which take place during the crawling of the cell over deformable elastic substrates. In addition, the kinematics of the migration under the effects of the varying degree of chemo-attractant concentration gradients will also be discussed. The trajectory of the cellīs center of mass, as well as cellsī polarization along gradient lines, will be shown to follow a quasi-periodic evolution with characteristic frequencies related to the biochemical processes regulating the internal remodeling of the cytoskeleton.
Scientific Computing Seminar
<--- 2006 Index