Statistical Inference for Nonlinear Functional Models with Application to Copy Number Variation and Multiple Myeloma Data

• Activity Number: 240
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Adrian Coles*+ and Arnab Maity and Veera Baladandayuthapani and Ganiraju Manyam
• Companies: North Carolina State University and North Carolina State University and The University of Texas MD Anderson Cancer Center and The University of Texas M.D. Anderson Cancer Center
• Keywords: Functional regression model ; Semiparametric regression model ; Linear mixed model ; Copy number ; Cytogenetics ; Biomarkers

Abstract

Recent research suggests that genomic copy number aberrations (CNAs) are associated with the development of cancer. The observed copy number profiles are often functional in nature due to the presence of serial correlation within the profiles. We are interested in identifying recurrent CNA regions that are associated with the biomarkers used to diagnose cancer, as well as estimating the effect of such regions on the biomarker. Many existing functional methods restrictively assume that the functional values are observed without error, assume a linear structure of the functional effect, and lack a procedure for testing the covariate effect. We propose a nonlinear functional regression model (NFRM) that models the copy number profiles non-parametrically. We develop estimation and testing procedures in this framework, while allowing for measurement error in the observed functional values. We show via simulations that NFRM performs significantly well when the relationship between the copy number profile and the biomarker is nonlinear but retains reasonable performance when the relationship is linear. We apply our method to a multiple myeloma data set.