Abstract
AbstractVarious gene network models with distinct physical nature have been widely used in biological studies. For temporal transcriptomic studies, the current dynamic models either ignore the temporal variation in the network structure or fail to scale up to a large number of genes due to severe computational bottlenecks and sample size limitation. On the other hand, correlation-based gene networks are more computationally more affordable, but have not been properly extended to gene expression time-course data.We propose Temporal Gene Coexpression Network (TGCN) for the transcriptomic time-course data. The mathematical nature of TGCN is the joint modeling of multiple covariance matrices across time points using a “low-rank plus sparse” framework, in which the network similarity across time points is explicitly modeled in the low-rank component. Using both simulations and a real data application, we showed that TGCN improved the covariance estimation loss and identified more robust and interpretable gene modules.
Publisher
Cold Spring Harbor Laboratory
Reference51 articles.
1. Maximum likelihood identification of Gaussian autoregressive moving average models
2. Akaike, H. (1998). Information theory and an extension of the maximum likelihood principle. In Selected Papers of Hirotugu Akaike, pages 199–213. Springer.
3. Azevedo, H. , Khaled, N. A. , Santos, P. , Bertonha, F. B. , and Moreira-Filho, C. A. (2017). Temporal analysis of hippocampal ca3 gene co-expression networks in a rat model of febrile seizures. Disease Models & Mechanisms.
4. Guidance for RNA-seq co-expression network construction and analysis: safety in numbers
5. Adaptive Thresholding for Sparse Covariance Matrix Estimation