Tracking of time-varying genomic regulatory networks with a LASSO-Kalman smoother

摘要: It is widely accepted that cellular requirements and environmental conditions dictate the architecture of genetic regulatory networks. Nonetheless, status quo in network modeling analysis assumes an invariant topology over time. In this paper, we refocus on a dynamic perspective networks, one can uncover substantial topological changes structure during biological processes such as developmental growth. We propose novel outlook inference time-varying from limited number noisy observations, by formulating estimation target tracking problem. overcome observations (small n large p problem) performing compressed domain. Assuming linear dynamics, derive LASSO-Kalman smoother, which recursively computes minimum mean-square sparse estimate connectivity at each time point. The LASSO operator, motivated sparsity allows simultaneous signal recovery compression, thereby reducing amount required observations. smoothing improves incorporating all track networks life cycle Drosophila melanogaster. recovered show few genes are permanent, whereas most transient, acting only specific phases organism.