March 2009
Issue #26
Home
Biomedical Research

Computational Systems Biology:
Understanding Biological Systems from
the Perspective of Networks and Dynamics

Xing-Ming Zhao1, Graziano Chesi2, and Luonan Chen1,3

1Institute of Systems Biology,
Shanghai University, 99 Shangda Road, Shanghai 200444, China
2Department of Electrical and Electronic Engineering,
University of Hong Kong, Pokfulam Road, Hong Kong
3Department of Electrical Engineering and Electronics,
Osaka Sangyo University, Nakagaito 3-1-1, Daito, Osaka 574-8530, Japan.

Email: , ,

Abstract

Systems Biology is undoubtedly a new field of primary interest worldwide that aims to understand and explain how living organisms achieve their complex biological functions through the quantitative analysis of networks of dynamically interacting biological components. From perspective of engineering, biological systems can be treated as complex systems, where various mathematical models and computer algorithms can be developed and utilized to understand the behavior and function of biological systems. This paper focuses on reviewing our recent works related to systems biology, especially focusing on aspects of dynamics and networks for biological systems. We present a general framework for modeling and analyzing the molecular network, and using the network for inferring regulatory relationship, signaling pathways, functional annotation, and so on.

1 Introduction

Systems biology is a recent emergent area in biology which focuses on the systematic study of complex interactions in biological systems by integrating mathematics, chemistry, physics, informatics, engineering and other fields. It has been found that a complicated living organism cannot be fully understood by merely analyzing individual components, such as genes and metabolites, where the interactions among components give rise to the function and behavior of the biological systems. Biologists face the challenge of understanding the complexities of living organisms. Instead of analyzing individual components or interactions of the organism with the so called reductionist approach, systems biology studies an organism by putting all the components and interactions together, and treats the organism as a dynamic and interacting network of genes, proteins and biochemical reactions which give rise to life. In recent years, with the rapid progress of biological science, many high-throughput technologies have been developed for systematically studying interactions or networks of molecules, such as microarray, the two-hybrid assay, co-immunoprecipitation and the ChIP-chip approach, which can be used to screen for protein-protein interaction (PPI) or to infer a gene regulatory network. With increasingly accumulated data from high-throughput technologies, molecular networks and their dynamics have been studied extensively from various aspects of living organisms. This research has helped biologists not only to understand complicated biochemical phenomena but also to elucidate the essential principles or fundamental mechanisms of cellular systems at the system-wide level. This article focuses on reviewing our recent work related to systems biology from the theoretical and engineering perspective, especially emphasizing the aspects of dynamics and networks for biological systems.

2 Dynamics

Dynamics exist in living organisms at all levels. From the perspective of both theoretical and experimental viewpoints, it is a big challenge in biological science to model, analyze and further predict the dynamic behaviors of biosystems. One of the best studied dynamics or rhythmic phenomena so far is circadian oscillations, which are assumed to be produced by limit cycle oscillators at the molecular level from the gene regulatory feedback loops. With the rapid advances in mathematics and experiments concerning the underlying regulatory mechanisms, more sophisticated theoretical models and general techniques are increasingly demanded to elucidate dynamic behaviors in a cell at a system-wide level. Closely related to systems biology, synthetic biology is also a new area of research that combines science and engineering in order to design and build novel biological functions and systems, and requires the techniques from systems biology. Research in synthetic biology is aimed at combining knowledge from various disciplines including molecular biology, engineering and mathematics to design and implement new cellular behaviors. Recent progress in genetic engineering has made the design and implementation of artificial synthetic gene networks realistic from both theoretical and experimental viewpoints. Actually, from the theoretical predictions, several simple gene networks have been experimentally constructed, e.g. genetic toggle switch and repressilator. Such simple models clearly represent a first step towards logical cellular control by manipulating and monitoring biological processes at the DNA level, which can not only be used as building blocks to synthesize the artificial biological systems, but also have great potential for biotechnological and therapeutic applications [1].


Fig. 1. Regulation relationship among components in one cell, communication among multi-cells, and synchronization induced by external stimuli.

Another common phenomena in biology is collective behaviors, which are essential coordinated responses resulting from an integrated exchange of information by cell communication in both prokaryotes and eukaryotes. The ability to cooperate or communicate between cells is an absolute requisite to ensure appropriate and robust coordination of cell activities at all levels of organisms under an uncertain environment. To understand the mechanism of cooperative behaviors (such as chemotaxis and quorum sensing) for molecules is an essential topic within systems biology, which requires both mathematical and biological knowledge and insight. Generally, cooperative behavior, such as intercellular communication is accomplished by transmitting individual cell reactions via intercellular signals to neighboring cells and further integration to generate a global cellular response at the level of molecules, tissues, organs and a body. However, all cell components exhibit intracellular noise owing to random births and deaths of individual molecules, and extracellular noise owing to environment fluctuations as shown in Figure 1. Gene regulation in particular, is an inherently noisy process which involves stochastic fluctuations owing to low copy numbers of many molecules per cell and uncertainty of an external environment. Such stochastic noise may not only affect the dynamics of the entire system but may also be exploited by living organisms to actively facilitate certain functions, such as synchronization and communication. In order to deal with this problem, we developed both numerical algorithms and analytical theory to analyze the cooperative behaviors of a population of individual cells induced by the noise by considering effects of stochastic fluctuations and signal diffusion processes.

In this area, we propose a general framework for modeling gene regulatory networks and cooperative behaviors, and develop theoretical techniques to design bio-molecular networks, which can be summarized as follows.

From the viewpoint of nonlinear dynamics, there are three major difficulties in analyzing a large-scale biological system, i.e. nonlinearity, noise and delays. In this work, we present a general framework to handle these problems by exploiting the special properties of cellular systems.

3 Molecular Networks and Interactions

The accumulation of large amount of ‘omics’ data provides an opportunity for mathematicians and computer scientists to develop new methods to infer molecular networks. For example, Large-scale microarray gene expression data and proteomics data provide new ways to construct gene regulatory networks [22, 23] or signaling pathways. Due to the complexity of data, an important challenging problem of reverse engineering is how to integrate various information sources for inferring a reliable molecular network, e.g. gene regulatory network. In contrast to the conventional methods that are mainly based on a single time-course dataset, we recently proposed a novel method GNRinfer[24] to infer a gene regulatory network which integrates multiple time-course microarray datasets even with different conditions in a universal framework. The method theoretically ensures the derivation of the most consistent network structure with respect to all of the datasets, thereby remarkably improving the prediction reliability. Actually, GNRinfer can be further extended to identify the conserved network patterns or motifs from the datasets of either the same species or different species. Furthermore, we proposed a novel method TRNinfer [25] to infer transcriptional regulatory networks (TRN) from gene expression data based on protein transcription complexes and mass action law, where the difference between GRN and TRN is that the interaction between Transcription factors (TFs) and target genes in TRN is physical interaction rather than indirect interaction in GRN. The proposed method not only can easily incorporate ChIP-Chip data as prior knowledge but also can integrate multiple gene expression datasets from different experiments simultaneously, thereby high prediction accuracy is expected. In addition, with integration of gene expression data and protein-protein interaction data, we proposed a novel linear programming (LP) model to infer signal transduction networks from protein-protein interaction network [26], where our method shows exciting results for identifying real signaling pathways. In practice, our LP model can be further applied to detect active pathways under various conditions, e.g. a specific stage of a disease. Most recently, we proposed a network flow model [27, 28] to identify signaling pathways from protein-protein interaction network, where the network flow model can guarantee a connected pathway identified from protein interaction network.


Fig.2 Inferring domain-domain interactions from protein-protein interactions

Recently, an interesting class of algorithms based on a statistical analysis was adopted to discover protein interactions at the domain level. With training data, those methods first calculate the probability of each domain pair to interact with each other, and then predict PPI based on the domain-domain interaction (DDI) information. Since DDI has a clear biological implication, it has been widely adopted to derive the PPI. In order to deal with this problem, we have recently developed, by exploiting special structures and composition of experimental data, a new Association Probabilistic Method (APM) [29] based on domain interaction to infer PPIs, which outperforms other existing methods in terms of prediction quality and computational efficiency. By taking into account multi-domain cooperations, we developed a new method to predict PPIs based on DDIs identified by considering multi-domain cooperations [30], where it is found that multi-domain cooperation indeed exists in protein interactions. Furthermore, we proposed a novel discriminative algorithm, namely domain interaction prediction with discriminative approach (DIPD) [31], to identify DDIs from PPIs as shown in Figure 2. The results from several benchmark data sets demonstrate that DIPD significantly outperforms existing generative methods, and predicted DDIs can be further used to predict PPIs and verify predicted PPIs by other approaches.

In biology, molecular networks orchestrate the sophistic and complex functions of the living cells. Various organisms differ not only because of differences in constituting proteins, but also because of the architectures of their molecular networks. Hence, it is essential to address the similarities and differences in the molecular networks by comparative network analysis, which can be directly applied to analyzing signaling pathways, conserving modules, discovering new biological functions or understanding the evolution of protein interactions. The problem of network alignment is to detect subnetworks that are conserved across species or within species by comparing two networks. Due to the high complexity in comparing such molecular networks, most of the conventional approaches either restrict comparative analysis to special structures, such as pathways without loops, or adopt heuristic algorithms due to computational burden. To overcome the difficulty of such computational complexity, we developed an alignment tool MNAligner [32, 33] based on an integer quadratic programming model to align networks in an accurate and efficient manner. The method is rather general and can be applied not only to unweighted and undirected networks, but also to weighted and directed networks [32, 33].

With various molecular networks available, it is possible to annotate uncharacterized components with the knowledge of known components according to their distribution across the network [34]. We proposed a novel method AGPS [35] to predict protein functions with integration of various data sources. In contrast to existing methods that consider proteins out of one functional family as negative samples, AGPS treated them as unlabeled data and achieved a higher prediction accuracy by generating a more reliable set of negative samples. Since protein domains are structural and functional units of proteins, it will be easier to infer the functions of proteins if we know their functions. Recently, we developed a new method to annotate domains based on the integration of various information including DDI network [36], and our method can annotate domains in a more accurate and reliable way compared to other methods. In addition, the proteins belonging to the same community are more likely have similar functions. We developed a new method to detect the community structure in the molecular network [37] and found some interesting hubs of network motifs in the PPI network [38]. The modular or motif underlying biological networks can provide insight into biomarker prediction. Recently, we proposed a new method to discover network biomarkers for major adverse cardiac events [39]. The network biomarkers are believed to be more robust than single biomarker found through gene differential expression analysis, where it is found that the biomarkers found for the same disease by different groups rarely overlap. The network biomarkers will provide insight into the mechanism underlying complex diseases and shed light on drug development.

4 Conclusions

One of the big challenges in systems biology is to build a complete and high-resolution description of molecular topography and connect molecular interactions with physiological responses. By studying the relationships and interactions between various parts of a biological system [40], e.g. metabolic pathways, organelles, cells, physiological systems and organisms, we aim eventually to develop an understandable model of the whole system, which is a key both for our? understanding of life and for the application of human medicine, in particular from the theoretical and engineering perspective. All software and related documents are available from http://www.isb.shu.edu.cn, http://intelligent.eic.osaka-sandai.ac.jp, or http://zhangroup.aporc.org/ResearchBioinformatics or upon request from authors. All the work listed in this article are collaborations with the research groups of Prof. Xiang-Sun Zhang (Chinese Academy of Sciences) and Prof. Kazuyuki Aihara (The University of Tokyo).

Acknowledgement

The research reported in the article was partially funded by the National High Technology Research and Development Program of China (2006AA02Z309), Innovation Funding of Shanghai University (A.10-0112-08-407), Key Project of Shanghai Education Committee and JSPS-NSFC collaboration project.

References

[1] P. Charusanti, X. Hu, L. Chen, D. Neuhauser, J. Distefano, “A mathematical model of BCR-ABL autophosphorylation, signaling through the CRKL pathway, and gleevec dynamics in chronic myeloid leukemia”, Discrete and Continuous Dynamical Systems - Series B, 4: 99-114, 2004.

[2] L. Chen, K. Aihara, “Stability of genetic regulatory networks with time delay”, IEEE Trans. on Circuits and Systems - I: Fundamental Theory and Applications, 49: 602-608, 2002.

[3] L. Chen, K. Aihara, “A model of periodic oscillation for genetic regulatory systems”, IEEE Trans. on Circuits and Systems - I: Fundamental Theory and Applications, 49: 1429-1436, 2002.

[4] C. Li, L. Chen, K. Aihara, “Stability of genetic networks with sum regulatory logic”, IEEE Trans. on Circuits and Systems - I: Fundamental Theory and Applications, 53: 2451-2458, 2006.

[5] G. Chesi and Y. S. Hung, “Stability analysis of uncertain genetic SUM regulatory networks”, Automatica, 44:9, pp. 2298-2305, 2008.

[6] L. Chen, R. Wang, T. Kobayashi, K. Aihara, “Dynamics of gene regulatory networks with cell division cycle”, Physical Review E, 70: 011909, 2004.

[7] T. Kobayashi, L. Chen, K. Aihara, “Modeling genetic switches with positive feedback loops”, Journal of Theoretical Biology, 221:379-399, 2002.

[8] R. Wang, Z. Jing, L. Chen, “Modelling periodic oscillation in gene regulatory networks by cyclic feedback systems”, Bulletin of Mathematical Biology, 67: 339-367, 2005.

[9] R. Wang, L. Chen, K. Aihara, “Construction of genetic oscillators with interlocked feedback networks”, Journal of Theoretical Biology, 242: 454-463, 2006.

[10] R. Wang, L. Chen, K. Aihara, “Detection of cellular rhythms and global stability within interlocked feedback systems”, Mathematical Biosciences, 209: 171-189, 2007.

[11] R. Wang, T. Zhou, Z. Jing, L. Chen, “Modelling periodic oscillation of biological systems with multiple time scale networks”, Systems Biology, 1:71-84, 2004.

[12] L. Chen, R. Wang, “Designing gene regulatory networks with specified functions”, IEEE Trans. on Circuits and Systems - I, 53: 2444-2450, 2006.

[13] R. Wang, L. Chen, “Synchronizing genetic oscillators by signaling molecules”, Journal of Biological Rhythms, 20: 257-269, 2005.

[14] T. Zhou, L. Chen, R. Wang, “A mechanism of synchronization in interacting multi-cell genetic systems”, Physica D, 211: 107-127, 2005.

[15] C. Li, L. Chen, K. Aihara, “Synchronization of coupled nonidentical genetic oscillators”, Physical Biology, 3: 37-44, 2006.

[16] C. Li, L. Chen, K. Aihara, “Stochastic synchronization of genetic oscillator networks”, BMC Systems Biology, 1: 6, 2007.

[17] C. Li, L. Chen, K. Aihara, “Transient Resetting: a novel mechanism for biological synchrony”, PLoS Computational Biology, 2: e103, 2006.

[18] R. Wang, L. Chen,K. Aihara, “Synchronizing a multicellular system by external input: an artificial control strategy”, Bioinformatics, 22: 1775 -1781, 2006.

[19] L. Chen, R. Wang, T. Zhou, K. Aihara, “Noise-induced cooperative behavior in a multi-cell system”, Bioinformatics, 21: 2722-2729, 2005.

[20] T. Zhou,L. Chen, K. Aihara, “Molecular communication through stochastic synchronization induced by extracellular fluctuations”, Physical Review Letter, 95: 178103, 2005.

[21] M. Springer, J. Paulsson, “Harmonies from Noise”, Nature, 439: 27-28, 2006.

[22] R.-S. Wang, X.-S. Zhang, L. Chen, “Inferring transcriptional interactions and regulator activities from experimental data”, Molecules and Cells, 24(3): 307-315, 2007.

[23] L.Chen, R.Wang, X-S.Zhang. Biomolecular Networks: Methods and Applications in Systems Biology. John Wiley & Sons, Inc., New Jersey, 2009.

[24] Y. Wang, T. Joshi, D. Xu, X-S. Zhang, L. Chen, “Inferring gene regulatory networks from multiple microarray datasets”, Bioinformatics, 22: 2413-2420, 2006.

[25] R.-S. Wang, Y. Wang, X.-S. Zhang, L. Chen, “Inferring transcriptional regulatory networks from high-throughput data”, Bioinformatics, 23: 3056-3064, 2007.

[26] X.-M. Zhao, R. Wang, L. Chen, K. Aihara, “Uncovering signal transduction networks from high-throughput data by integer linear programming”, Nucleic Acids Research, 39(9):e48, 2008.

[27] X.-M. Zhao, R. Wang, L. Chen, K. Aihara, “Automatic modeling of signaling pathways by network flow model”, Journal of Bioinformatics and Computational Biology, in press, 2009.

[28] Z.Wu, X.-M. Zhao, L. Chen, “Identifying responsive functional modules from protein-protein interaction network”, Molecules and Cells, 27(3), in press, 2009.

[29] L. Chen, L-Y. Wu, Y. Wang, X-S. Zhang, “Inferring protein interactions from experimental data by association probabilistic method”, Proteins, 62:833-837, 2006.

[30] R. Wang, Y. Wang, L.-Y. Wu, X.-S. Zhang, L. Chen, “Analysis on multidomain cooperation for predicting protein-protein interactions”, BMC Bioinformatics, 8:391, 2007.

[31] X.-M. Zhao, L. Chen, “Domain-domain interaction identification with a feature selection approach”, Lecture Notes in Computer Science, 5265: 178-186, 2008.

[32] Z. Li, S. Zhang, X. -S. Zhang, L. Chen, “Alignment of protein interaction networks by integer quadratic programming”, IEEE EMBC06: 5527- 5530, 2006.

[33] Z. Li, S. Zhang, Y. Wang, X-S. Zhang, L. Chen, “Alignment of molecular networks by integer quadratic programming”, Bioinformatics, 23: 1631-1639, 2007.

[34] X.-M. Zhao, L.Chen, K. Aihara, “Protein function prediction with high-throughput data”, Amino Acids, 35:517-530, 2008.

[35] X.-M. Zhao, Y. Wang, L. Chen, K. Aihara, “Gene function prediction using labeled and unlabeled data”, BMC Bioinformatics, 9:57, 2008.

[36] X.-M. Zhao, Y. Wang, L. Chen, K. Aihara, “Protein domain annotation with integration of heterogeneous information sources”, Proteins: Structure, Function, and Bioinformatics, 72(1): 461 - 473, 2008.

[37] R.-S. Wang, Y. Wang, X.-S. Zhang, L. “Chen,Detecting community structure in complex networks by optimal rearrangement clustering”, Lecture Notes in Artificial Intelligence, 4819:119-130, 2007.

[38] G. Jin, S. Zhang, X.-S. Zhang, L. Chen, “Hubs with network motifs organize modularity dynamically in the protein-protein interaction network of yeast”, PLoS ONE, 2(11): e1207, 2007.

[39] G. Jin, X. Zhou, H. Wang, H. Zhao, K. Cui, X. Zhang, L. Chen, S. Hazen, K. Li, S. Wong, “The knowledge-integrated network biomarkers discovery for Major Adverse Cardiac Events”, Journal of Proteome Research, 7(9): 4013-4021, 2008.

[40] S. Zhang, G. Jin, X-S. Zhang, L. Chen, “Discovering functions and revealing mechanisms at molecular level from biological networks”, Proteomics, 7: 2856-2869, 2007.

About the Authors

Xingming Zhao received the B.E. and M.E. degrees from Jilin University, China, in 2000 and 2003, respectively. He received his Ph.D. degree from University of Science and Technology of China, China, in 2005. From May 2006 to May 2008, he is a researcher in ERATO Aihara Complexity Modelling Project, JST. He is currently an associate professor of Institute of Systems Biology, Shanghai University, Shanghai, China.

Graziano Chesi received the Laurea degree in Information Engineering from the University of Firenze in 1997 and the Ph.D. in Systems Engineering from the University of Bologna in 2001. He was a visiting scientist at the University of Cambridge (1999-2000) and he held fellowships from the Japanese Society for the Promotion of Science and the European Union at the University of Tokyo (2001-2004). He was with the University of Siena (2000-2006) and he joined the University of Hong Kong in 2006. Dr. Chesi was the recipient of the Best Student Award of the Faculty of Engineering of the University of Firenze in 1997. He is an Associate Editor of Automatica, an Associate Editor of the IEEE Transactions on Automatic Control, and a Guest Editor of the Special Issue on ''Positive Polynomials in Control'' of the IEEE Transactions on Automatic Control. His research interests include computer vision, nonlinear systems, robotics, robust control, and systems biology.

Luonan Chen received the M.E. and Ph.D. degrees in electrical engineering from Tohoku University, Sendai, Japan, in 1988 and 1991, respectively. Since 1997, he has been a member of the Faculty of Osaka Sangyo University, Osaka, Japan, where he is currently a Professor in the Department of Electrical Engineering and Electronics. He is the founding director of Institute of Systems Biology, Shanghai University. His fields of interest are systems biology, bioinformatics, and nonlinear dynamics. Dr. Chen is a Senor IEEE member. He serves as an editor and editorial board member for many international journals related to systems biology, e.g. Associate Editor at BMC Systems Biology, and Associate Editor at IEEE/ACM Transactions on Computational Biology and Bioinformatics. In recent six years, he published over 70 journal papers related to systems biology.