For survival data with high-dimensional covariates, results generated in the analysis of a single dataset are often unsatisfactory because of the small sample size. Integrative analysis pools raw data from multiple independent studies with comparable
designs, effectively increases sample size, and has better performance than meta-analysis and single-dataset analysis. In this study, we conduct integrative analysis of survival data under the accelerated failure time (AFT) model. The sparsity structures of multiple datasets are described using the homogeneity and heterogeneity models. For variable selection under the homogeneity model, we adopt group penalization approaches. For variable selection under the heterogeneity model, we use composite penalization and sparse group penalization approaches. As a major advancement from the existing studies, the asymptotic selection and estimation properties are rigorously established. Simulation study is conducted to compare different penalization methods and against alternatives. We also analyze four lung cancer prognosis datasets with gene expression measurements.
In genome-wide association studies (GWAS), penalization is an important approach for identifying genetic markers associated with trait while mixed model is successful in accounting for a complicated dependence structure among samples. Therefore, pena
lized linear mixed model is a tool that combines the advantages of penalization approach and linear mixed model. In this study, a GWAS with multiple highly correlated traits is analyzed. For GWAS with multiple quantitative traits that are highly correlated, the analysis using traits marginally inevitably lose some essential information among multiple traits. We propose a penalized-MTMM, a penalized multivariate linear mixed model that allows both the within-trait and between-trait variance components simultaneously for multiple traits. The proposed penalized-MTMM estimates variance components using an AI-REML method and conducts variable selection and point estimation simultaneously using group MCP and sparse group MCP. Best linear unbiased predictor (BLUP) is used to find predictive values and the Pearsons correlations between predictive values and their corresponding observations are used to evaluate prediction performance. Both prediction and selection performance of the proposed approach and its comparison with the uni-trait penalized-LMM are evaluated through simulation studies. We apply the proposed approach to a GWAS data from Genetic Analysis Workshop (GAW) 18.
Precision is a virtue throughout science in general and in optics in particular where carefully fabricated nanometer-scale devices hold great promise for both classical and quantum photonics [1-6]. In such nanostructures, unavoidable imperfections of
ten impose severe performance limits but, in certain cases, disorder may enable new functionalities [7]. Here we demonstrate on-chip random nanolasers where the cavity feedback is provided by the intrinsic disorder in a semiconductor photonic-crystal waveguide, leading to Anderson localization of light [8]. This enables highly efficient and broadband tunable lasers with very small mode volumes. We observe an intriguing interplay between gain, dispersion-controlled slow light, and disorder, which determines the cross-over from ballistic transport to Anderson localization. Such a behavior is a unique feature of non-conservative random media that enables the demonstration of all-optical control of random lasing. Our statistical analysis shows a way towards ultimate thresholdless random nanolasers.
