2010/07/03 by Ali Shojaie, George Michailidis, Shojaie, Ali +1 · 3 citations
Biochemistry, Genetics and Molecular Biology · Mathematics · #Gene expression and cancer classification #Statistical Methods and Inference #Gene Regulatory Network Analysis
paper · pdf · doi:10.48550/arxiv.1007.0499
Components of biological systems interact with each other in order to carry\nout vital cell functions. Such information can be used to improve estimation\nand inference, and to obtain better insights into the underlying cellular\nmechanisms. Discovering regulatory interactions among genes is therefore an\nimportant problem in systems biology. Whole-genome expression data over time\nprovides an opportunity to determine how the expression levels of genes are\naffected by changes in transcription levels of other genes, and can therefore\nbe used to discover regulatory interactions among genes.\n In this paper, we propose a novel penalization method, called truncating\nlasso, for estimation of causal relationships from time-course gene expression\ndata. The proposed penalty can correctly determine the order of the underlying\ntime series, and improves the performance of the lasso-type estimators.\nMoreover, the resulting estimate provides information on the time lag between\nactivation of transcription factors and their effects on regulated genes. We\nprovide an efficient algorithm for estimation of model parameters, and show\nthat the proposed method can consistently discover causal relationships in the\nlarge p, small n setting. The performance of the proposed model is\nevaluated favorably in simulated, as well as real, data examples. The proposed\ntruncating lasso method is implemented in the R-package grangerTlasso and is\navailable at http://www.stat.lsa.umich.edu/~shojaie.\n