A Bayesian Hierarchical Model for High-Dimensional Meta-analysis
Many biomedical applications are concerned with the problem of selecting important predictors from a high-dimensional set of candidates, with the gene expression data as one example. Due to the fact that the sample size in any single study is usually small, it is thus important to combine information from multiple studies. In this chapter, we introduce a Bayesian hierarchical modeling approach which models study-to-study heterogeneity explicitly to borrow strength across studies. Using a carefully formulated prior specification, we develop a fast approach to predictor selection and shrinkage estimation for high-dimensional predictors. The proposed approach, which is related to the relevance vector machine (RVM), relies on maximum a posteriori (MAP) estimation to rapidly obtain a sparse estimate. As for the typical RVM, there is an intrinsic thresholding property in which unimportant predictors tend to have their coefficients shrunk to zero. The method will be illustrated with an application of selecting genes as predictors of time to an event.
Key wordsBayesian hierarchical model MAP estimation meta analysis relevance vector machine shrinkage
This research was supported in part by the Statistical and Applied Mathematical Sciences Institute (SAMSI) Summer 2008 research program on Meta-analysis: Synthesis and Appraisal of Multiple Sources of Empirical Evidence. The gene barcode data used in this paper were kindly provided by Dr. Rafael Irizarry and Dr. Michael Zilliox. Research of Fei Liu was partially supported by the University of Missouri-Columbia research board award.
- 1.Ji, S., Dunson, D., and Carin, L. (2009) Multi-task compressive sensing. IEEE Transactions Signal Processing 57, 92–106.Google Scholar
- 3.Ip, E. H. S. (1994) A stochastic EM estimator in the presence of missing data: theory and applications. Ph.D. thesis, Department of Statistics, Stanford University.Google Scholar
- 9.Miller, M., Wang, C., Parisini, E., Coletta, R., Goto, R., Lee, S., Barral, D., Townes, M., Roura-Mir, C., Ford, H., Brenner, M., and Dascher, C. C.(2005) Characterization of two avian MHC-like genes reveals an ancient origin of the cd1 family. Processings of National Academy of Science, USA, 102, 8674–8679.Google Scholar
- 10.Pawitan, Y., Bjohle, J., Amler, L., Borg, A., Egyhazi, S., Hall1, P., Han, X., Holmberg, L., Huang, F., Klaar, S., Liu, E., Miller, L., Nordgren, H., Ploner, A., Sandelin, K., Shaw, P., Smeds, J., Skoog, L., Wedren, S., and Bergh, J. (2005) Gene expression profiling spares early breast cancer patients from adjuvant therapy: derived and validated in two population-based cohorts. Breast Cancer Research 7, R953–R964.PubMedCrossRefGoogle Scholar
- 11.Sotiriou, C., Wirapati, P., Loi, S. A. H., Fox, S., Smeds, J., Nordgren, H., Farmer, P., Praz, V., Haibe-Kains, B. C. D., Larsimont, D., Cardoso, F., Peterse, H., Nuyten, D. M. B., Van de Vijver, M. J. B., Piccart, M., and Delorenzi, M. (2006) Gene expression profiling in breast cancer: Understanding the molecular basis of histologic grade to improve prognosis. Journal of National Cancer Institute 98, 262–272.CrossRefGoogle Scholar