American Statistical Association
New York City
Metropolitan Area Chapter

New York State Psychiatric Institute
at Columbia University Medical Center
Biostatistics Seminar



CONSTRUCTING HIGH-RESOLUTION CANDIDATE
GENE REGIONS FROM DENSE SNP LINKAGE DATA:
A SUBSAMPLING BASED APPROACH

by

William C.L. Stewart, Ph.D.
Department of Biostatistics
Columbia University


Abstract

Complex traits in human genetics are multifactorial; and, identifying the precise location of even a single disease susceptibility (DS) gene is an arduous task. The process often begins with the identification of the correct chromosome, and typically, the next step is to localize the DS gene relative to a panel of densely spaced single nucleotide polymorphism (SNPs). However, the genotypes at nearby SNPs are often correlated and, in practice, DS gene localization based on maximum likelihood (ML) estimation is computationally infeasible. As a result, existing methods either (1) ignore the correlation; (2) approximate the correlation; or, (3) reduce the correlation by restricting attention to genotypes at sparsely spaced SNPs.

We propose a consistent and asymptotically normal estimator of DS gene location that averages location estimates across random subsamples of the original dense SNP linkage data. Within each subsample, the SNPs are highly polymorphic and approximately uncorrelated, which permits quick and efficient computation of the ML estimator for each subsample. From the analysis of simulated data, we show that existing approaches can yield biased and/or inefficient estimation. In particular, for the simulation scenarios that we consider, the results show that our estimator has the following: smaller variance than approach (3); either squared bias or variance less than or equal to approach (2); and, when combined with the nonparametric bootstrap procedure, yields an asymptotically valid, high-resolution confidence interval for the true DS gene location. Consequently, our confidence interval can be used to define high-resolution candidate gene regions, and we expect that it will be an important tool in the multi-staged process of identifying the precise location of DS genes.

Biographical Note

William C. L. Stewart is currently Assistant Professor of Biostatistics with internal funding for projects that connect genetic variation (both structural and allelic) to the phenotypic variation of complex traits. He received his Ph.D. at the University of Washington in Statistics, with an emphasis in Statistical Genetics. Dr. Stewart is a member of the American Society of Human Genetics, the International Society of Epidemiology, and has published in several interdisciplinary journals, including Biometrics, Genome Research, and Genetic Epidemiology. His current applied projects include studies for hypertension, epilepsy, and bipolar disorder; his current methodological work involves the development of improved methods for mapping disease susceptibility genes, and for the analysis of copy number variation.


Date: Tuesday, March 10, 2009
Time: 3:00 - 4:00 P.M.
Location: New York State Psychiatric Institute
1051 Riverside Drive
6th Floor Multipurpose Room (6602)
New York, New York
(Directions)

RESERVATIONS ARE NOT REQUIRED

Coffee: 2:45 to 3:00 P.M.
Reception: 4:00 to 4:30 P.M.


Home Page | Chapter News | Chapter Officers | Chapter Events
Other Metro Area Events | ASA National Home Page | Links To Other Websites
NYC ASA Chapter Constitution | NYC ASA Chapter By-Laws

Copyright © 2009 by New York City Metropolitan Area Chapter of the ASA
Designed and maintained by Cynthia Scherer
Send questions or comments to nycasa@mindspring.com