Seminars
To receive Biostatistics seminar email announcements, send an email to LISTSERV@LISTSRV.UCSF.EDU, with blank subject and a one-line message: subscribe BIOSTATEVENTS firstname lastname.
Seminars marked "Research Program" are informal discussions of research by UCSF faculty. For more information, contact John Neuhaus.
The CAPS Methods Core Quantitative Methods Working Group also presents seminars of interest to statisticians.
February 15, 2012
Athanasios Kottas, PhD
Department of Applied Mathematics and Statistics
University of California Santa Cruz
BAYESIAN NONPARAMETRIC MIXTURE MODELING FOR DEVELOPMENTAL TOXICOLOGY DATA
We present a Bayesian nonparametric modeling framework for replicated count responses in dose-response experiments. The focus will be on modeling and risk assessment in developmental toxicity studies, where the primary objective is to determine the relationship between the level of exposure to a toxic chemical and the probability of a physiological or biochemical response. Data from these experiments comprise clustered categorical responses and typically involve features that can not be captured by standard parametric approaches. To provide flexibility in the functional form of both the response distribution and the dose-response curves, the proposed mixture models are built from dependent Dirichlet process priors, with the dependence of the mixing distributions governed by the dose level. The practical utility of the methodology is illustrated with data from a toxicity experiment for which the dose-response curves for different endpoints have different shapes, including a non-monotonic, possibly hormetic, dose-response relationship.
Joint work with Kassandra Fronczyk, Post-doctoral Researcher, Department of Biostatistics, The University of Texas MD Anderson Cancer Center.
February 1, 2012
Daniel Stram, PhD
Department of Preventive Medicine, University of Southern California
ESTIMATING AND INTERPRETING HERITABILITY FROM GENOME WIDE ASSOCIATION STUDIES
Many studies have investigated the heritability of complex traits in genome wide association studies (GWAS). To date the top associations (i.e. those judged globally significant after correction for multiple comparisons) for most complex traits only seem to explain a relatively small fraction of observed trait heritability, as estimated in family studies. Various interpretations of these findings are possible: there is intense current interest, for example, in determining the role that rare variation may play in the genetic architecture of these traits. Others, for example Zuk et al (PNAS 2012) argue that narrow- sense additive heritability has been over-estimated for many traits and that simple polygenic approaches (summing the effects of many variants with no epistatic contribution) cannot be expected to explain either individual phenotype variation or trait resemblance between close relatives. However other analyses of complex traits, e.g. those of Yang et al (Nat Genet, 2010, 2011), find a signal of strong additive heritability in GWAS data that involve common variants, even though identification of the particular common variants that make contributions to this strong polygenic signal is not yet possible (presumably because the signal of each is very small). An extreme version of this is embodied in the analyses of Purcell et al (Nature 2009) in which polygenic scores involving half of all markers examined appeared to be predictive of schizophrenia and bipolar disorder. In my talk I will review these arguments, discussion the estimation of additive heritability in GWAS data using variance components methods and discuss ongoing analyses of the heritability of three phenotypes: height, prostate cancer, and breast cancer, using GWAS studies taking place within a multiethnic cohort.
January 25, 2012
6th Annual Special Invited Seminar
PLEASE NOTE
Time: 3:00 - 4:00 p.m. followed by Social Hour
**Please note different location - Radiology Classroom 331, Lobby 6
Every year the Division of Biostatistics at UCSF invites one researcher of national eminence for a special lecture. This year we are pleased to have Dr. Nan Laird from Harvard University.
Host: Joan Hilton, ScD, (joan@biostat.ucsf.edu)
Nan Laird is the Harvey V. Fineberg Professor of Biostatistics at the Harvard School of Public Health. Dr. Laird has contributed to methodology in many different fields, including meta-analysis, statistical genetics, and longitudinal data. She is a coauthor of the book, Applied Longitudinal Analysis, with Garrett Fitzmaurice and James Ware and is a coauthor the book, Fundamentals of Modern Statistical Genetics, with Christoph Lange. She is the recipient of many awards and prizes, including Fellow of the American Statistical Association, the Samuel Wilks Award in 2011.
Nan Laird, PhD
Department of Biostatistics
Harvard University
ANALYSIS OF SEQUENCE DATA WITH FAMILIES
Over the past few years, association analysis has become the primary tool for finding genes that underlie complex traits. Both population- based and family-based designs are commonly used designs in genetic association studies. Recent technological advances in exome and genome sequencing are designed to make the next-generation of sequence-based association studies affordable. We review some general approaches to sequence analysis, then discuss a novel method for family designs.
Family-based designs are completely robust to population substructure, a useful feature for association studies with rare variants. Moreover our proposed method can have better power than the weighted sum approach, one of the first methods proposed for analyzing rare variants in a case-control design.
