Large Causal Gene Regulatory Network Inference by Decomposition into Subnetworks

Inferring the gene regulatory network is an important first step toward understanding the work of the cell and consequently curing diseases related to malfunction of the cell. One thorny problem in gene regulatory network inference is that even with high throughput technology, the available time series expression data is still very limited compared to the network size. To alleviate this problem, we propose to decompose large network into small subnetworks without prior knowledge of the decomposition. Our algorithm first infers an initial GRN using CLINDE, decomposes it into possibly overlapping subnetworks, then infers each subnetwork by either CLINDE or DD-lasso and finally merges the subnetworks. We have tested this algorithm on synthetic data of networks with 500 and 1000 genes. The results show that our proposed algorithm does improve the GRN inference performance of using either CLINDE or DD-lasso alone on the large network, with statistical significance, and is robust to different variances and slight deviation from Gaussian distribution in error terms.