Sig2GRN: A Software Tool Linking Signaling Pathway with Gene Regulatory Network for Dynamic Simulation Authors: Fan Zhang, Runsheng Liu and Jie Zheng Presented by: Fan Wu School of Computer Science and Engineering, Nanyang Technological University, Singapore Oct-2016
Outline Background and Motivation Methods and Implementation Results Discussions and Conclusions 2
Background One of the major forms of cellular responses to extracellular perturbations is to change the gene expression in response to the cellular signals transmitted by signaling pathways. Stimuli cellular signals transcription factor (TF) activities subsequent cellular behaviors gene expression patterns Many studies have presented various computational strategies, such as data-driven, logic-based and biochemical kinetic methods, in modeling signaling pathways or gene regulatory networks separately. 3
Motivation Linking computational models of signaling pathways to predicted cellular responses such as gene expression regulation is a major challenge in computational systems biology. Here, we plan to integrate signal transduction with gene expression regulation so that time-course gene expression data can be simulated given the user-defined external stimuli to the signaling pathways. 4
Outline Background and Motivation Methods and Implementation Results Discussions and Conclusions 5
The overall strategy 6
Modeling of signal transduction Generalized logical modeling of signaling pathways for predicting transcription factor activities X " = 1 d X "'( + 1 * 1 A, * 1 B. 1 X "'( * 1 A, 1 * 1 B. X "'(,.,. In this formula, X " is the activity level (value [0,1]) of node x at time t. d is a pre-defined parameter denoting the degradation rate. A, (or B. ) is the signals, i.e., the product of the activity level and the edge weight (value [0,1]), transmitted from the i-th activating (or j-th inhibiting) parent node. The updated state (value [0,1]) is defined by the nondegraded part and the newly activated part minus the inhibited part. F. Zhang and H. Chen and L.N. Zhao and H. Liu and T.M. Przytycka and J. Zheng. Generalized logical model based on network topology to capture the dynamical trends of cellular signaling pathways. BMC Systems Biology, 10(Suppl 1):7, 2016. 7
Modeling of transcriptional regulation Boolean model AND logical gate for TFs with same regulation type. Inhibition is assumed to precede the activation. A gene will be switched ON (or OFF) when the maximum activity level of activating (or inhibiting) transcription factors surpasses a pre-defined threshold. Thermodynamics-based the gene expression level is defined as a function of the activity levels of the bound transcription factors [E] is the gene expression level, N is the number of all possible arrangements of TFs, G is the set of transcription factor arrangements that turn the gene on, n i (n m ) is the number of TFs employed in the i-th (m-th) arrangement, K j and [TF j ] represent the binding affinity of binding site j and the activity level of the TF corresponding to site j, Q i is the probability of a gene being expressed in the i-th arrangement. 8
Outline Background and Motivation Methods and Implementation Results Discussions and Conclusions 9
Case study 1: DNA damage induced cell apoptosis The network of cell apoptosis regulation induced by DNA damage. Simulation steps: Select receptors (ATM, ATR and DNA-PK), and set input levels. Set edge weights. Generate the dynamics of the nodes based on the generalized logical model. Select the TFs that regulate the gene of interest. For example, for gene Bcl-2, select E2F1 and p53. Choose either Boolean or Thermodynamics-based model. Generate the time-course gene expression patterns. 10
Case study 1: DNA damage induced cell apoptosis Simulation settings: ATR and DNA-PK are selected as input nodes to simulate the exposure of the cells to UV light, and the input levels are both 1. The edge weights of activation and inhibition interactions are 0.7 and 0.8, respectively. For Bax, the selected TF is p53; for Bcl-2, the TFs are E2F1 and p53. 11
Case study 1: DNA damage induced cell apoptosis Simulation outputs and validation: 12
Case study 2: apoptotic signaling network treated by different combinations of drugs The network of cell fates (e.g., apoptosis, proliferation and cell cycle) regulation. Simulation settings (four treatments): The input level: The edge weights of activation and inhibition interactions are 0.7 and 0.8, respectively. 13
Case study 2: apoptotic signaling network treated by different combinations of drugs Simulation results: The network does not involve transcriptional regulation, the predicted dynamics of Caspase 3 (the only upstream node of Apoptosis) is considered as the predicted cell responses to the perturbations. Validation: The dataset has no measurement of gene expression, instead, the numbers of cells that fall into each cell fate were measured at 5 time points (i.e., 0, 6, 8, 12 and 24 hours). Therefore, we directly calculate the proportion of the dead cells at each time point as the cells response to the perturbations. 14
Outline Background and Motivation Methods and Implementation Results Discussions and Conclusions 15
Discussions and Conclusions Computational simulation is an important systems biology approach to the analysis of signaling pathways and gene regulatory networks. In this work, we present Sig2GRN, a parameter-free software, which is able to link the cellular signaling pathways with the downstream gene expression regulation. Generalized logical modeling of signal transduction Boolean or thermodynamical modeling of gene expression regulation. Sig2GRN is extensible so that more computational models of gene regulation can be integrated to facilitates studies in systems biology. Future improvements Reveal the synergistic effect of the co-treatment by multiple drugs. Incorporate multiscale modeling and simulation. For example, to deal with the issue on mapping the simulation iterations to the real time points. 16
Acknowledgements This project is supported by MOE AcRF Tier 2 Grant ARC39/13 (MOE2013-T2-1- 079) and MOE AcRF Tier 1 seed grant on complexity (RGC2/13), Ministry of Education, Singapore. We would like to thank Ms. Jing Guo, a Ph.D. student at the School of Computer Science and Engineering, Nanyang Technological University, for her help with testing the software. Thank you Software availability: http://histone.scse.ntu.edu.sg/sig2grn/ 17