Both gene expression levels (GEs) and copy number alterations (CNAs) have important implications in the development of complex diseases. GEs are partly regulated by CNAs, and much effort has been devoted to understanding their relations. The expression of a gene can be regulated by multiple CNAs, and one CNA can regulate the expression of multiple genes. In addition, multiple GEs (CNAs) can be correlated with each other. The existing methods for associating GEs with CNAs have limitations in deciphering the complex data structures. In this study, we develop a sparse double Laplacian shrinkage approach. It jointly models the effects of multiple CNAs on multiple GEs. Penalization is adopted to identify sparse and interpretable associations. Network adjacency is computed to describe the interconnections among GEs (CNAs). Two Laplacian shrinkage penalties are introduced to accommodate the network adjacency in estimation. Simulation shows that the proposed approach outperforms the benchmark with more accurate marker identification. With TCGA data on gliobalstoma (GBM), we analyze GEs and CNAs in the apoptosis pathway, and demonstrate advantages of the proposed method.
- Genetic regulation,
- High-dimensional data,
- Regularized estimation
Available at: http://works.bepress.com/shuangge/49/