BMC Medical Genomics · 2017 · 17 citations · 32 references
Life SciencesBioinformaticsCell BiologyComputational BiologyComputational Systems BiologyGene Regulatory NetworksGene Expression DynamicsQuantitative Cell BiologyComputational Biology ModelsGaussian Process RegressionGene Expression Dynamics Along TrajectoriesGene Network DynamicsGene Regulatory Network ModelsRegulatory Gene NetworksTemporal Gene Expression Dynamics
BACKGROUND: Recent advances in omics technologies have raised great opportunities to study large-scale regulatory networks inside the cell. In addition, single-cell experiments have measured the gene and protein activities in a large number of cells under the same experimental conditions. However, a significant challenge in computational biology and bioinformatics is how to derive quantitative information from the single-cell observations and how to develop sophisticated mathematical models to describe the dynamic properties of regulatory networks using the derived quantitative information. METHODS: This work designs an integrated approach to reverse-engineer gene networks for regulating early blood development based on singel-cell experimental observations. The wanderlust algorithm is initially used to develop the pseudo-trajectory for the activities of a number of genes. Since the gene expression data in the developed pseudo-trajectory show large fluctuations, we then use Gaussian process regression methods to smooth the gene express data in order to obtain pseudo-trajectories with much less fluctuations. The proposed integrated framework consists of both bioinformatics algorithms to reconstruct the regulatory network and mathematical models using differential equations to describe the dynamics of gene expression. RESULTS: The developed approach is applied to study the network regulating early blood cell development. A graphic model is constructed for a regulatory network with forty genes and a dynamic model using differential equations is developed for a network of nine genes. Numerical results suggests that the proposed model is able to match experimental data very well. We also examine the networks with more regulatory relations and numerical results show that more regulations may exist. We test the possibility of auto-regulation but numerical simulations do not support the positive auto-regulation. In addition, robustness is used as an importantly additional criterion to select candidate networks. CONCLUSION: The research results in this work shows that the developed approach is an efficient and effective method to reverse-engineer gene networks using single-cell experimental observations.
32
Inferring Regulatory Networks from Expression Data Using Tree-Based Methods
Vân Anh Huynh‐Thu, Alexandre Irrthum, Louis Wehenkel et al. · PLoS ONE · 2010 · 2.3K citations · Full text
Yuichi Taniguchi, Paul Choi, Gene‐Wei Li et al. · Science · 2010 · 2.1K citations · Full text
Wisdom of crowds for robust gene network inference
Daniel Marbach, James C. Costello, Robert Küffner et al. · Nature Methods · 2012 · 1.8K citations · Full text