|
Abstract:
|
Advances in high-throughput biotechnologies have provided an unprecedented opportunity for biomarker discovery, which, from a statistical point of view, can be cast as a variable selection problem.This problem is challenging due to the high-dimensional and non-linear nature of omics data, and it generally suffers three difficulties, an unknown functional form, variable selection consistency, and highly-demanding computation. To circumvent theses difficulties, we employ a feed-forward neural network to approximate the unknown nonlinear function and then conduct structure selection for the neural network by choosing appropriate prior distributions that lead to the consistency of variable selection. We propose to resolve the computational issue by implementing the population stochastic approximation Monte Carlo algorithm on the OpenMP platform. The numerical results indicate that the proposed method can execute veryfast on a multicore computer and work very well for identification of relevant variables for general high-dimensional nonlinear systems. The proposed method is successfully applied to selection of anticancer drug response genes for the cancer cell line encyclopedia (CCLE).
|
