Analysis of Information Flows in Interaction Networks: Implication for Drug Discovery and Pharmacological Research
Abstract: Frequent failures of experimental medicines in clinical trials question current concepts for predicting drug-effects in the human body. Improving the probability for success in drug discovery requires a better understanding of cause-effect relationships at the organism, organ, tissue, cellular, and molecular levels, each having a different degree of complexity. Despite the longstanding realization that clinical and preclinical drug-effect information needs to be integrated for generating more accurate forecasts of drug-effects, a road map for linking these disparate sources of information currently does not exist. This review focuses on a possible approach for obtaining these relationships by analyzing causes and effects on the basis of the topology of network interaction systems that process information at the cellular and organ system levels.
During the past century, pharmacological and medical research has yielded an enormous amount of information linking drug effects observed at a molecular or cellular level to effects observed in the human body (Whittaker, 2003). The resulting knowledge has profoundly shaped the architectures of drug discovery platforms and has focused the search for the treatment of complex diseases on the identification of small molecules capable of affecting functions of specific cellular machinery (Colquhoun, 2006). Unfortunately, molecular structure-centered approaches to drug discovery frequently fail to predict the spectrum of diverse effects generated by medicines in the human body. In fact, the high failure rate of experimental medicines in clinical trials combined with the associated financial burden questions the economic viability of this approach for the discovery of new medicines (Eichler et al., 2010). Improving probabilities for success requires capabilities that more accurately predict cause-effect relationships at the organism, organ, tissue, cellular, and molecular levels (Wist et al., 2009; Noble, 2010; Szalma et al., 2010). Since each of these systems has different scales and structural complexities, the establishment of comprehensive cause-effect relationships requires understanding of the functions and properties of dynamic interaction-network systems which process information at these different levels (Bassingthwaighte et al., 2006; Gunawardena, 2010; Molina et al., 2010). This review focuses on methodology for creating these links by investigating information transfers between dynamic interaction-network systems. Herein we provide examples of applications of this approach in medicine and pharmacological research (Werner, 2005; Lim et al., 2006; Fliri et al., 2010).
Focusing on the Information Flow in Cause-Effect Relationship Analysis
Recent advances in proteomics have increased our knowledge of how medicines affect biological systems by identifying thousands of cellular pathways and networks describing millions of dynamic interactions (Figure 1) between proteins, genes, and small molecules (see http://pathguide.org/). Analysis of this large body of molecular interaction information suggests that a much more complex regulatory scheme needs to be considered for connecting drug-induced cellular causes with effects in the whole organism than is suggested by the currently accepted and relied upon methodology. In this respect, traditional cause-effect analysis identifies relationships between the structures of medicines and the generation of effects caused by modulation of discrete functions of biochemical pathways containing specific drug targets (Jørgensen and Linding, 2010). Breaking away from this reductionistic “hard wiring” concept, an emerging systems perspective proposes that the functions of biochemical pathways do not operate in isolation but instead are integrated into much larger inter- and intra-cellular information processing network systems that are regulated by the information flow induced by environmental changes, administration of medicines, and disease causing events (Tkacik et al., 2008; Shiraishi et al., 2010; de Ronde et al., 2010; Terry et al., 2007; Missiuro et al., 2009). An important aspect of this dynamic perspective is that the information flow itself becomes the central regulatory scheme synchronizing system-wide functions. Thus, it is increasingly recognized that large biological interaction network systems do not only conduct signal-induced information flows but they also process information. This capacity to process information is based on the modulation of network topologies resulting from the dynamic making and breaking of connections between different network components (Nakano et al., 2007; Perkins et al., 2009; Bowsher, 2011). This dynamic modulation of interaction network topologies by information flows gives rise to emergent distributed computations (George et al., 2003). Examples of these molecular interaction-based computation systems have been described in plants, bacteria, and higher organisms (Mehta et al., 2009; Mott and Peak, 2007; Gong and van Leeuwen, 2009; Emmert-Streib and Dehmer, 2009). Hence, identification of cause-effect relationships in complex biological systems requires not only information on the structures of interacting network components but also information on network topologies conducting information flows and carrying out emergent distributed computations (Crutchfield and Mitchell, 1995; Bouchard and Osbourn, 2006; Muller et al., 2008). In this respect, the routing of medication-induced information flows in cellular protein networks provides the computational scheme regulating induction of organ system network topologies which, in turn, synchronize functions of organ systems generating drug effects and/or disease symptoms (Poudret et al., 2008; Shiraishi et al., 2010). Currently, only the interactions between proteins are partially known. Since cellular information processing that determines cause-effect relationships in organisms is also mediated by interactions of proteins with DNA or RNA and between RNA molecules, the capacity of probabilistic models for ascertaining network topologies involved in cause-effect relationships of medicines or disease is limited and will be limited until this information gap is resolved (Gilad et al., 2008). In contrast, information flow analysis-based approaches, focused on in this review, do not suffer from this limitation (Aldridge et al., 2009; Fliri et al., 2009).
Determining Interaction Network Topologies
Rationalizing drug actions or disease states using dynamic interaction network perspectives poses major challenges. Generation of a disease must induce stable network topologies wherein transitions between stable states reflect stages of disease progression (Shiraishi et al., 2010). Conversely, pharmacotherapy seeks to destabilize disease network topologies and ideally reverse the process of disease progression by inducing network topologies resembling the normal regulatory stable state. In this respect, network-based, cause-effect analysis of disease and pharmacotherapy seeks to gain an understanding of circuitry regulating stability of network topologies associated with disease and targeted by pharmacotherapy (Yeger-Lotem and Margalit, 2003; Kitano, 2007). The robustness of the networks and consequently their resilience to perturbations seem to be regulated by network topology and information flows (Shiraishi et al., 2010).
Hence, most methods for determining interaction network topologies involved in cause-effect analysis of medicines or disease start out by determining the distribution of signal-induced information flows in cellular, trans-cellular, or organ-network systems (Lim et al., 2006; Ciaccio et al., 2010; Fliri et al., 2007). The methodology available for this purpose has its roots in modern communication network systems, which determine network topologies involved in the dynamic routing of information transfers by sending messages (tracer probes) from a source to a destination in order to identify network positions that are reached during message transfers (Tian and Shen, 2006). In this context, the term “network reachability information” introduced in communication networks relates to changes in measurements of observations at discrete network nodes, caused by the sending of messages from a source to a destination. Application of this approach in a pharmacological context equates the “sending of a message” to the administration of a medicine or the introduction of disease-causing events and concomitant protein network perturbations (also frequently referred to as input signals) and “message destination” equates to drug-effects or disease symptoms observed at the organism level (output).
For identifying sources and targets in biological network systems, methods such as expression quantitative trait loci (eQTL) mapping experiments are frequently employed, even though there are many other means as well (Gilad et al., 2008; Fliri et al., 2010; Fliri et al., 2005b). Moreover, in lieu of tracer probes, medicines, molecular probes, point mutations in proteins, or even introduction of disease-causing events can be used for initiating information flows starting from ascertainable protein network positions (Fliri et al., 2007; Kreeger and Lauffenburger, 2010; Lim et al., 2006). Since it is not practical to monitor all possible network positions for determining the distribution of the information flow triggered by medication or disease, most methods used for this purpose preselect certain network-nodes as monitoring points (Figure 2) (Viswanathan et al., 2008). Methods for determining the distribution of information flows in cellular protein networks, induced by drugs or disease-causing events include, amongst other methods, quantitative mass spectrometry (de Godoy et al., 2006), phosphopeptide enrichment methods (Macek et al., 2009), aptamer technology (Radko et al., 2007), micro-western arrays (Ciaccio et al., 2010), or even text mining (Fliri et al., 2007; Campillos et al., 2008). Likewise, distribution of information flows, induced by drugs or disease-causing events in organ system networks, can be determined by identifying specific organ functions affected by medication or disease (Takarabe et al., 2008; Campillos et al., 2008; Ren et al., 2010). If network nodes interacting with the pre-selected “monitoring positions” are known, routes of the drug or disease-induced information flow in protein-networks can be inferred by linking network nodes reached by a signal-induced information flow with proteins that are capable of interacting with the “monitoring positions.” The information required for this network-connectivity analysis is usually obtained from independent protein interaction databases such as, for example, KEGG (http://www.genome.jp/kegg), BioCarta (http://www.biocarta.com), MetaCore http://www.genego.com), and Ingenuity (http://www.ingenuity.com). If reliable protein interaction information cannot be found in these databases, protein interaction information can be obtained by using, for example, yeast-2-hybrid experiments or imaging methods (Nicholson et al., 2007).
Investigations of System Cause-Effect Relationships
Making determination of interaction network topologies exceedingly complex is the fact that many cellular constituents involved in the regulation of dynamic interaction network topologies have more than one interaction partner (Figure 3) which, in turn, may have more than one function (Zhou et al., 2010).
Since the dynamic breaking and making of links in interaction networks is, in principle, capable of affecting the information exchange between all other network components, dynamic variation of network topologies enables efficient distribution of locally generated information throughout the entire network system (Kumar et al., 2008). While these dynamic network designs provide a robust regulatory scheme, they are nevertheless susceptible to even minor topology perturbations. For example, investigations on cellular effects associated with mutations in RAS proteins indicate that even small changes in local interaction network topologies can be translated into broad reaching systems effects (Kreeger and Lauffenburger, 2010). The example provided by point-mutated RAS proteins illustrates the main challenge faced by probability-based network topology analysis since even minor differences in network linkage may be associated with major changes of in vivo effects. In recognition of this problem, many different methods have been developed for constructing interaction network topologies. Most of these methods rely on the identification of alignments between experimental observations and known protein-protein interaction information (Stegle et al., 2010; Raman, 2010; Aldridge et al., 2009; Jiang and Singh, 2010). For example, Bayesian statistics and hidden Markov models have been used to construct network topology-based, cause-effect relationship models (Janes and Lauffenburger, 2006). One limitation of current methodology employed for constructing network topologies is the frequent reliance on the assumption that the fastest route for transmitting information in biological communication network systems involves information transfer between nearest network neighbors (Sethi et al., 2009; Kuchaiev et al., 2010). This connectivity mode may not always be correct because connections between network nodes that belong to different circuits, but have closely related functions, cannot be excluded in information-processing networks (Pan et al., 2010; Freeman, 1997). However, these connections likely play a role in translating protein network topology information into predictions of drug-induced physiological effects (Araujo et al., 2007; Dixon and Stockwell, 2009; Mani et al., 2008; Lehár et al., 2009a; Shiraishi et al., 2010). The advantage of focusing on information flow in system-wide, interaction-network topology analysis is that, in principle, no detailed knowledge is required on interacting network components or interaction kinetics (Aldridge et al., 2009; Ding et al., 2006; Ren et al., 2010). Using information flow analysis in combination with fuzzy logic-based, network topology modeling approaches enables identification of complex systems cause-effect relationships even though not all components or kinetic parameters regulating the topologies of pertinent interaction networks are known (Aldridge et al., 2009; Fliri et al., 2009). Obviously, reliance on any network-topology framework in pharmacological or medical research will always require experiments for selecting from a nearly infinite number of similar interaction network topologies, those topologies that best rationalize all pharmacology and pathology observations (Muller et al., 2008; Molina et al., 2010; Morris et al., 2010).
Methods for investigating cause-effect relationships in complex biological systems frequently rely on Granger causality and use of “input” and “output” information transfer functions (Blinowska. et al., 2004; Ding et al., 2006; Aldridge et al., 2009; Fliri et al., 2007). This approach anticipates that the induction of specific protein interaction network topologies, which conducts the drug- or disease-induced information flow throughout cellular network-systems, precedes the induction of organ system network topologies generating drug- or disease-induced effect patterns in organisms (Tkacik et al., 2008; Fliri et al., 2007; Russell and Aloy, 2008). Pertinent “output” information transfer functions describing generation of effects in organisms are obtained, for example, by ascertaining drug- or disease-induced changes in functions of different organ systems (Fliri et al., 2007; Campillos et al., 2008). Likewise “input” information transfer functions capturing drug effects at cellular levels are obtained, for example, by determining drug-induced modifications in either structure or functions of proteins, genes or RNA in preselected monitoring positions (Lim et al., 2006; Mitsos et al., 2009). For correlating network topologies associated with input and output information transfer functions, several different techniques have been used (Aldridge et al., 2009; Fliri et al., 2007). For example, association rules derived from hierarchical clustering of input and output information transfer functions enable interaction network topology construction and simultaneous analysis of information flows induced by many different medicines. This approach provides a global perspective on network topology alignments governing cause-effect relationships triggered by many different system perturbations (Fliri et al., 2007; Aldridge et al., 2009).
Using Network Topologies for Understanding the Pharmacology of Medicines
The translation of cellular pharmacology information into organisms’ responses to drug treatment depends on many factors influenced by chemistry and biology (Hu and Agarwal, 2009). For instance, drugs with high target specificity will not achieve their anticipated therapeutic objective (Figure 4) if the cellular machinery targeted by selective drug design exchanges information with other cellular machines that execute different and important cellular functions (Li et al., 2008; Kumar et al., 2008; Wallach et al., 2010).
Thus, drug-effect predictions, derived from analysis of cellular pathway connectivity containing the drug target, will not be able to forecast the generation of pleiotropic drug effects if the drug target-containing pathway exchanges information with other pathways (Xie et al., 2009). Likewise, information transfer between different pathways may lead to the generation of similar pleiotropic drug effects even though medicines might target different proteins in different pathways (Fliri et al., 2009). Hence, one application of protein network topology-based, cause-effect analysis is in creating an understanding of the role of cross pathway communication in generating pleiotropic drug effects (Li et al., 2008; Donaldson and Calder, 2010). For example, by using correlations between side effect profiles and protein interaction data, network models for drug-induced, cross-pathway communication have been proposed (Wallach et al., 2010; Fliri et al., 2010).
Information Flow Analysis in Network-based, Cause-Effect Analysis of Human Disease
Use of information flow analysis in investigating origins of human disease is illustrated in an investigation examining relationships between proteins causing the inherited form of the coordination disorder, ataxia (Lim et al., 2006). Starting with observations indicating that mutations in 36 different proteins cause inheritable forms of ataxia in humans and mice, the cDNAs of these disease-causing proteins were used as bait in yeast-2-hybrid experiments for identifying interaction partners of these proteins (Molina et al., 2010). Analysis of the obtained protein interaction information identified one large interconnected network system which used, as bridges for the information transfer, 541 interaction partners enabling information exchange between the 36 ataxia-associated proteins. For identifying cellular functions regulated by the ataxia network topology, gene ontology term (GO term) analysis was applied (Young et al., 2005). This analysis suggests that the ataxia network topology regulates functions of different pathways such as, for example, transcription factor binding, transcription cofactor activity, DNA binding, and inhibition of kinase activity. Inspection of the ataxia network topology suggests that the information transfer between different pathways is mediated by hub proteins (proteins capable of interacting with multiple partners) (Yu et al., 2007). From a cause-effect perspective, this study suggests that perturbation of the information flow conducted by this regulatory sub network would lead to abnormal regulation of cellular function and concomitant expression of characteristic symptom patterns at the organism level (Fliri et al., 2005a). Similar cause-effect analysis linking disease-causing events and expression of characteristic symptoms patterns have been described for asthma, schizophrenia, Alzheimer’s, inflammatory diseases, and cardiovascular disease. The summary of pertinent findings suggests that disease-causing proteins affect the information exchange between proteins linked through discrete interaction network topologies (Barrenas et al., 2009; Wallach et al., 2010). For determining these cause-effect relationships, investigators frequently rely on genetic information. In this respect, genetic approaches provide protein-interaction information that can be directly linked to the expression of characteristic disease symptoms patterns. However, while this linkage is ideal for conducting meaningful network topology-based, cause-effect analysis, genetic methodology may not be ideal for investigations conducted during drug discovery because of its low throughput and the expense of generating fusion proteins.
An alternative methodology for obtaining protein-interaction information for cause-effect analysis is described in an investigation involving ligands for the epidermal growth factor receptor (EGFR) (Mitsos et al., 2009). Breaking with the dogma that cause-effect relationships should focus on the affinity of ligands for their putative drug targets, the authors of this study set out to identify protein network nodes that can be reached by EGFR ligand-induced information flow by monitoring phosphorylation events at key positions of the EGFR pathway. Although the information flow analysis conducted in this study is pathway centered, it nevertheless enabled differentiation of EGFR pathway topologies induced by different ligands. Moreover, this investigation identified phosphorylation events that cannot be explained by inspecting the affinities of these ligands for respective drug targets (Mitsos et al., 2009).
An advantage of pathway-centered methodology is that protein network construction can benefit from the abundance of known pathway information, which assists in the selection of appropriate monitoring positions for information flow analysis. Not surprisingly, this strategy has not only found application for cause-effect analysis but also in the identification of drug targets. An example of this application is provided by an investigation involving a network model of Erb signaling and its use in developing novel antibody-based treatments (Schoeberl et al., 2009). In considering approaches like this in drug discovery settings, one should be aware of its limitations. The most important limitation of pathway-focused methodology is that unanticipated cross pathway communication, which could significantly impact cause-effect analysis, cannot be detected outside of a broader network perspective. Second, it is not always clear how drug- or disease-induced information flows are routed through cellular network systems and how they impact the functions of different pathways (Borisov et al., 2009). Third, while a specific protein might be an attractive target in the context of a linear pathway, the same protein may no longer be a good target from a systems perspective (Muller et al., 2008).
For comprehensive assessment of network topologies regulating cross pathway communication, live imaging methods have been developed that provide unprecedented topological details in information flow analysis (Nicholson et al., 2007). Unfortunately, from a drug screening perspective, this method may be less desirable than pathway-centered methods because throughput of live imaging is generally low and live imaging requires a large number of experiments to cover all molecules of interest (Cheong et al., 2009). An alternative approach, quantitative mass spectrometry, offers to overcome this throughput bottleneck and enables large-scale measurements using different cell populations and treatment conditions, making this methodology potentially a versatile drug-screening tool (de Godoy et al., 2006; Cox and Mann, 2007). Moreover, recent advances in phosphopeptide enrichment methods enable simultaneous detection and quantization of thousands of phosphorylation sites and, as a result, these methods are likely to become more frequently used in system-wide, information flow analysis (de Godoy et al., 2006; Macek et al., 2009; Olsen et al., 2006).
Additional sources for network reachability information are accessible by using chemogenomics methods. These in silico approaches are based on the anticipation that capability of inducing protein network topology perturbations is linked to the affinity of drugs for various protein scaffolds. For example, this approach was used for estimating protein-network reachability information associated with the pharmacology of various cholesteryl ester transfer protein (CETP) inhibitors. By characterizing the structure of the putative ligand binding site of known CETP inhibitors, the authors identified proteins with similar ligand binding characteristics (Xiao et al., 2009). Using the in silico predicted protein network reachability information for different CETP inhibitors and known pathway information, likely interaction partners for the putative network reachable proteins were identified. The comparison of protein network topologies induced by different CETP inhibitors was used for identifying differences in cross pathway communication and in vivo pharmacology. Noteworthy, JTT-705, the most promiscuous of the three highlighted CETP inhibitors, did not cause hypertension as a side effect while the more selective Torcetrapib did, suggesting that adverse events of medicines may not necessarily be reduced by increasing a drug’s target selectivity but rather by fine tuning pertinent cross pathway communication (Funder, 2009).
Aligning Protein Interaction and Drug-Effect Network Topologies of Medicines
Considering the complexity of dynamic regulatory network topologies, success of using protein interaction networks in drug discovery will depend on the ability of aligning protein interaction network topologies with drug-effect information obtained from in vitro and in vivo measurements (Sanderson, 2009; Campillos et al., 2008). Development of this capability is described in a recent multi-system analysis involving 1,300 medicines (Fliri et al., 2009). This investigation used hierarchical clustering of information spectra derived through text mining for identifying protein network reachability information associated with medicines generating similar drug-effect profiles (Fliri et al., 2005a). The typical scaling problem encountered in multi-system, cause-effect analysis was overcome by correlating the topologies of drug-effect networks with the topologies of cellular protein interaction networks (Ding et al., 2006; Aldridge et al., 2009; Fliri et al., 2007). These topology comparisons, in turn, were achieved by using, in lieu of network topologies, the underlying network reachability information and by quantifying similarities between normalized information spectra (Bassingthwaighte et al., 2006; Fliri et al., 2007). This approach identified quantitative correlations between pleiotropic in vivo drug-effect similarities and protein network topologies conducting drug-induced information flows through cellular network systems. Using Rosiglitazone and Glimepiride as illustrative examples, this investigation indicates that medicines with different mechanisms of action will produce similar in vivo effects if the protein network topologies induced by them cause similar cross pathway communication. Importantly, this result suggests that, for drug-effect predictions, knowing the protein network topology induced by drugs is more valuable than knowing the point of entry (drug target).
Using a broad range of heterogeneous protein interaction and drug-effect data in information flow analysis (IFA) provides a new avenue for investigating cause-effect relationships in drug discovery (Tarassov et al., 2008; Shiraishi et al., 2010; Zhang and Kast, 2010; Missiuro et al., 2010). In pharmacological settings, the IFA approach can take advantage of diverse sets of known chemical probes, natural products, and effect databases (Campillos et al., 2008; Fliri et al., 2007). Using cause-effect analysis as basis for constructing interaction network topologies, in combination with predictive modeling and carefully designed experiments, provides a roadmap for identifying circuits regulating transitions between various protein network topologies involved in pharmacological outcomes and disease progression (Wallach et al., 2010; Raman, 2010; Platts et al., 2010). Based on these combinatory analyses, the fine tuning of regulatory network topologies may well become a feasible alternative to molecular structure-based drug discovery and provide superior medicines and drug combinations (Lehár et al., 2007; Lehár et al., 2009b; Feala et al., 2010; Yang et al., 2008). Finally, as the detail and accuracy of disease- and drug-induced network topologies improve, so too will our understanding of biological systems (Pujol et al., 2010). This improved understanding will likely impact drug target selections and the success rate of new medicines (Han, 2008; Lusis and Weiss, 2010).
Anton F. Fliri, Ph.D., is the Chief Executive Officer of SystaMedic Inc., 1084 Shennecossett Road, Groton, Connecticut 06340, USA. William T. Loging, Ph.D., is the Director of Computational Biology, Boehringer-Ingelheim Pharmaceuticals Inc., 900 Ridgebury Road, Ridgefield, Connecticut 06877, USA. Robert A. Volkmann, Ph.D., is the Chief Scientific Officer of SystaMedic Inc., 1084 Shennecossett Road, Groton, Connecticut 06340, USA.
Robert A. Volkmann, Ph.D., Chief Scientific Officer, SystaMedic Inc., 1084 Shennecossett Road, Groton, Connecticut 06340, USA.
Araujo RP, Liotta LA, Petricoin EF. Proteins, drug targets and the mechanisms they control: the simple truth about complex networks. Nat Rev Drug Discov 6(11):871-880, 2007.
Barrenas F, Chavali S, Holme P, Mobini R, Benson M. Network properties of complex human disease genes identified through genome-wide association studies. PLoS One 4(11):e8090, 2009.
Bassingthwaighte JB, Chizeck HJ,Atlas LE.Strategies and tactics in multiscale modeling of cell-to-organ systems. Proc IEEE 94(4):819-830, 2006.
Blinowska KJ, Kus R, Kaminski M. Granger causality and information flow in multivariate processes. Phys Rev E 70(5):050902(R), 2004.
Borisov N, Aksamitiene E, Kiyatkin A, Legewie S, Berkhout J, Maiwald T, Kaimachnikov NP, Timmer J, Hoek JB, Kholodenko BN. Systems-level interactions between insulin-EGF networks amplify mitogenic signaling. Mol Syst Biol 5:256, 2009.
Bouchard AM, Osbourn GC. Dynamic self-assembly in living systems as computation. Natural Computing 5(4):321-362, 2006.
Bowsher, CG. Information processing by biochemical networks: a dynamic approach J R Soc Interface 8(55):186-200, 2011.
Campillos M, Kuhn M, Gavin AC, Jensen LJ, Bork P. Drug target identification using side-effect similarity. Science 321(5886):263-266, 2008.
Cheong R, Wang CJ, Levchenko A. High content cell screening in a microfluidic device. Mol Cell Proteomics 8(3):433-442, 2009.
Ciaccio MF, Wagner JP, Chuu C, Lauffenburger DA, Jones RB. Systems analysis of EGF receptor signaling dynamics with microwestern arrays. Nat Methods 7(2):148-155, 2010.
Colquhoun D. The quantitative analysis of drug-receptor interactions: a short history. Trends Pharm Sci 27(3):149-157, 2006.
Cox J, Mann M. Is proteomics the new genomics? Cell 130(3):395-398, 2007.
Crutchfield JP, Mitchell M. The evolution of emergent computation. Proc Natl Acad Sci U S A 92(23):10742-10746, 1995.
de Godoy LMF, Olsen JV, de Souza GA, Li G, Mortensen P, Mann M. Status of complete proteome analysis by mass spectrometry: SILAC labeled yeast as a model system. Genome Biol 7(6):R50, 2006.
de Ronde W, Tostevin F, ten Wolde PR. The effect of feedback on the fidelity of information transmission of time-varying signals. Phys Rev E Stat Nonlin Soft Matter Phys 82(3 Pt 1):031914, 2010.
Ding M, Chen Y, Bressler SL. Granger causality: Basic theory and application to neuroscience. In: Handbook of Time Series Analysis. Schelter BWM, Timmer J. (eds.). pp451-474. Wiley-VCH Verlage, Weinheim, Germany, 2006.
Dixon SJ, Stockwell BR. Identifying druggable disease-modifying gene products. Curr Opin Chem Biol 13(5-6):549-555, 2009.
Donaldson R, Calder M. Modelling and analysis of biochemical signalling pathway cross-talk. EPTCS 19:40-54, 2010.
Eichler HG, Aronsson B, Abadie E, Salmonson T. New drug approval success rate in Europe in 2009. Nat Rev Drug Discov 9(5):355-356, 2010.
Emmert-Streib F, Dehmer M. Information processing in the transcriptional regulatory network of yeast: Functional robustness. BMC Syst Biol 3:35, 2009.
Feala JD, Cortes J, Duxbury PM, Piermarocchi C, McCulloch AD, Paternostro G. Systems approaches and algorithms for discovery of combinatorial therapies. Wiley Interdisciplinary Reviews: Systems Biology and Medicine 2(2):181-193, 2010.
Fliri AF, Loging WT, Thadeio PF, Volkmann RA. Analysis of drug-induced effect patterns to link structure and side effects of medicines. Nat Chem Biol 1(7):389-397, 2005a.
Fliri AF, Loging WT, Thadeio PF, Volkmann RA. Biological spectra analysis: Linking biological activity profiles to molecular structure. Proc Natl Acad Sci U S A 102(2):261-266, 2005b.
Fliri AF, Loging WT, Volkmann RA. Analysis of system structure-function relationships. ChemMedChem 2(12):1774-1782, 2007.
Fliri AF, Loging WT, Volkmann RA. Drug effects viewed from a signal transduction network perspective. J Med Chem 52(24):8038-8046, 2009.
Fliri AF, Loging WT, Volkmann RA. Cause-effect relationships in medicine: a protein network perspective. Trends Pharm Sci 31(11):547-555, 2010.
Freeman LC. A set of measures of centrality based upon betweenness. Sociometry 40(1):35-41, 1997.
Funder J. The off-target effects of torcetrapib. Endocrinology 150(5):2024-2026, 2009.
Gilad Y, Rifkin SA, Pritchard JK. Revealing the architecture of gene regulation: the promise of eQTL studies. Trends Genet 24(8):408-415, 2008.
George S, Evans D, Marchette S. A biological programming model for self-healing. In: First ACM Workshop on Survivable and Self-Regenerative Systems. pp72-81. Association for Computing Machinery, New York, New York, USA, 2003.
Gong P, van Leeuwen C. Distributed dynamical computation in neural circuits with propagating coherent activity patterns. PLoS Comput Biol 5(12):e1000611, 2009.
Gunawardena J. Biological systems theory. Science 328(5978):581-582, 2010.
Han J-DJ. Understanding biological functions through molecular networks. Cell Res 18(2):224-237, 2008.
Hu G, Agarwal P. Human disease-drug network based on genomic expression profiles. PLoS One 4(8):e6536, 2009.
Janes KA, Lauffenburger DA. A biological approach to computational models of proteomic networks. Curr Opin Chem Biol 10(1):73-80, 2006.
Jiang P, Singh M. SPICi: a fast clustering algorithm for large biological networks. Bioinformatics 26(8):1105-1111, 2010.
Jørgensen C, Linding R. Simplistic pathways or complex networks? Curr Opin Genet Dev 20(1):15-22, 2010.
Kitano H. A robustness-based approach to systems-oriented drug design. Nat Rev Drug Discov 6(3):202-210, 2007.
Kuchaiev O, Milenkovic T, Milenković T, Memišević V, Hayes W, Pržulj N. Topological network alignment uncovers biological function and phylogeny. J R Soc Interface 7(50):1341-1354, 2010.
Kumar N, Afeyan R, Kim H-D, Lauffenburger DA. Multipathway model enables prediction of kinase inhibitor cross-talk effects on migration of HER2-overexpressing mammary epithelial cells. Mol Pharmacol 73(6):1668-1678, 2008.
Lehár J, Krueger AS, Avery W, Heilbut AM, Johansen LM, Price ER, Rickles RJ, Short III GF, Staunton JE, Jin X, Lee MS, Zimmermann GR, Borisy AA. Synergistic drug combinations tend to improve therapeutically relevant selectivity. Nat Biotechnol 27(7):659-666, 2009a.
Lehár J, Krueger AS, Zimmermann GR, Borisy AA. Therapeutic selectivity and the multi-node drug target. Discov Med 8(43):185-190, 2009b.
Lehár J, Zimmermann GR, Krueger AS, Molnar RA, Ledell JT, Heilbut AM, Short III GF, Giusti LC, Nolan GP, Magid OA, Lee MS, Borisy AA, Stockwell BR, Curtis T, Keith CT. Chemical combination effects predict connectivity in biological systems. Mol Syst Biol 3:80, 2007.
Li Y, Agarwal P, Rajagopalan D. A global pathway crosstalk network. Bioinformatics 24(12):1442-1447, 2008.
Lim J, Hao T, Shaw C, Patel AJ, Szabo G, Rual JF, Fisk CJ, Li N, Smolyar A, Hill DE, Barabási AL, Vidal M, Zoghbi HY. A protein-protein interaction network for human inherited ataxias and disorders of Purkinje cell degeneration. Cell 125(4):801-814, 2006.
Lusis AJ, Weiss JN. Cardiovascular networks: systems-based approaches to cardiovascular disease. Circulation 121(1):157-170, 2010.
Macek B, Mann M, Olsen JV. Global and site-specific quantitative phosphoproteomics: principles and applications. Annu Rev Pharmacol Toxicol 49:199-221, 2009.
Mani KM, Lefebvre C, Wang K, Lim WK, Basso K, Dalla-Favera R, Califano A. A systems biology approach to prediction of oncogenes and molecular perturbation targets in B-cell lymphomas. Mol Syst Biol 4:169, 2008
Mehta P, Goya S, Long T, Bassler BL, Wingreen NS. Information processing and signal integration in bacterial quorum sensing. Mol Syst Biol 5:325, 2009.
Missiuro PV, Liu K, Zou L, Ross BC, Zhao G, Liu JS, Ge H. Information flow analysis of interactome networks. PLoS Comput Biol 5(4):e1000350, 2009.
Mitsos A, Melas IN, Siminelakis P, Chairakaki AD, Saez-Rodriguez J, Alexopoulos LG. Identifying drug effects via pathway alterations using an integer linear programming optimization formulation on phosphoproteomic data. PLoS Comput Biol 5(12):e1000591, 2009.
Molina F, Dehmer M, Perco P, Graber A, Girolami M, Spasovski G, Schanstra JP, Vlahou A. Systems biology: opening new avenues in clinical research. Nephrol Dial Transplant 25:1015-1018, 2010.
Morris MK, Saez-Rodriguez J, Sorger PK, Lauffenburger DA. Logic-based models for the analysis of cell signaling networks. Biochemistry 49(15):3216-3224, 2010.
Mott KA, Peak D. Stomatal patchiness and task-performing networks. Ann Bot 99(2):219-226, 2007.
Muller M, Obeyesekere M, Mills GB, Ram PT. Network topology determines dynamics of the mammalian MAPK1,2 signaling network: bifan motif regulation of C-Raf and B-Raf isoforms by FGFR and MC1R. FASEB J 22(5):1393-1403, 2008.
Nakano T, Suda T, Koujin T, Haraguchi T, Hiraoka Y. Molecular communication through gap junction channels: System design, experiments and modeling. In: Proceedings of the 2nd International Conference on Bio-Inspired Models of Network, Information, and Computing Systems (BIONETICS 2007). Institute for Computer Sciences, Social-Information, and Telecommunications Engineering, Budapest, Hungary, 2007.
Nicholson RL, Welch M, Ladlow M, Spring DR. Small-molecule screening: advances in microarraying and cell-imaging technologies. ACS Chem Biol 2(1):24-30, 2007.
Noble D. Biophysics and systems biology. Phil Trans R Soc A. 368:1125-1139, 2010.
Olsen JV, Blagoev B, Gnad F, Macek B, Kumar C, Mortensen P, Mann M. Global, in vivo, and site-specific phosphorylation dynamics in signaling networks. Cell 127(3):635-648, 2006.
Pan RK, Chatterjee N, Sinha S. Mesoscopic organization reveals the constraints governing Caenorhabditis elegans nervous system. PLoS ONE 5(2):e9240, 2010.
Perkins TJ, Swain PS. Strategies for cellular decision-making. Mol Syst Biol 5:326, 2009.
Platts AE, Lalancette C, Emery BR, Carrell DT, Krawetz SA. Disease progression and solid tumor survival: a transcriptome decoherence model. Mol Cell Probes (24)1:53-60, 2010.
Poudret M, Arnould A, Comet J-P, Le Gall P, Meseure P, Képès F. Topology-based abstraction of complex biological systems: Application to the Golgi apparatus. Theory Biosci 127(2):79-88, 2008.
Pujol A, Mosca M, Farrés J, Aloy P. Unveiling the role of network and systems biology in drug discovery. Trends Pharm Sci 31(3):115-123, 2010.
Radko, SP, Rakhmetova SY, Bodoev NV, Archakov AI. Aptamers as affinity reagents for clinical proteomics. Biochemistry (Moscow) Supplement. Series B, Biomedical Chemistry 1(3):198-209, 2007.
Raman K. Construction and analysis of protein-protein interaction networks. Autom Exp 2(1):2, 2010.
Ren X, Zhou X, Wu L-Y and Zhang X-S. An information-flow-based model with dissipation saturation and direction for active pathway inference. BMC Syst Biol 4:72, 2010.
Russell RB, Aloy P. Targeting and tinkering with interaction networks. Nat Chem Biol 4(11):666-673, 2008.
Sanderson CM. The Cartographers toolbox: building bigger and better human protein interaction networks. Brief Funct Genomic Proteomic 8(1):1-11, 2009.
Schoeberl B, Pace EA, Fitzgerald JB, Harms BD, Xu L, Nie L, Linggi B, Kalra A, Paragas V, Bukhalid R, Grantcharova V, Kohli N, West KA, Leszczyniecka M, Feldhaus MJ, Kudla AJ, Nielsen UB. Therapeutically targeting erbB3: a key node in ligand-induced activation of the erbB receptor-PI3K axis. Sci Signal 2(77):ra31, 2009.
Sethi A, Eargle J, Black AA, Luthey-Schulten Z. Dynamical networks in tRNA:protein complexes. Proc Natl Acad Sci U S A 106(16):6620-6625, 2009.
Shiraishi T, Matsuyama S, Kitano H. Large-scale analysis of network bistability for human cancers. PLoS Comput Biol 6(7):e1000851, 2010.
Stegle O, Parts L, Durbin R, Winn J. A Bayesian framework to account for complex non-genetic factors in gene expression levels greatly increases power in eQTL studies. PLoS Comput Biol 6(5):e1000770, 2010.
Szalma S, Koka V, Khasanova T, Perakslis ED. Effective knowledge management in translational medicine. J Transl Med 8:68, 2010.
Takarabe M, Okuda S, Itoh M, Tokimatsu T, Goto S, Kanehisa M. Network analysis of adverse drug interactions. Genome Inform 20:252-259, 2008.
Terry LJ, Shows EB, Wente SR. Crossing the nuclear envelope: hierarchical regulation of nucleocytoplasmic transport. Science 318(5855):1412-1416, 2007.
Taniguchi CM, Emanuelli B, Kahn CR. Critical nodes in signalling pathways: insights into insulin action. Nat Rev Mol Cell Biol 7:85-96, 2006.
Tarassov K, Messier V, Landry CR, Radinovic S, Molina MMS, Shames I, Malitskaya Y, Vogel J, Bussey H, Michnick SW. An in vivo map of the yeast protein interactome. Science 320(5882):1465-1470, 2008.
Tian H, Shen H. An improved algorithm for multicast topology discovery from end-to-end measurements. Int J Communication Systems 19(8):935-953, 2006.
Tkacik G, Callan CG Jr, Bialek W. Information flow and optimization in transcriptional regulation. Proc Natl Acad Sci U S A 105(34):12265-12270, 2008.
Viswanathan GA, Seto J, Patil S, Nudelman G, Sealfon SC. Getting started in biological pathway construction and analysis. PLoS Comput Biol 4(2):e16, 2008.
Wallach I, Jaitly N, Lilien R. A structure-based approach for mapping adverse drug reactions to the perturbation of underlying biological pathways. PLoS One 5(8):e12063, 2010.
Werner E. Genome semantics, in silico multicellular systems and the Central Dogma. FEBS Lett 579(8):1779-1782, 2005.
Whittaker PA. What is the relevance of bioinformatics to pharmacology? Trends Pharm Sci 24(8):434-439, 2003.
Wist AD, Berger SI, Iyengar R. Systems pharmacology and genome medicine: a future perspective. Genome Med 1:11, 2009.
Xie L, Li J, Xie L, Bourne PE. Drug discovery using chemical systems biology: identification of the protein-ligand binding network to explain the side effects of CETP inhibitors. PLoS Comput Biol 5(5):e1000387, 2009.
Yang K, Bai H, Ouyang Q, Lai L, Tang C. Finding multiple target optimal intervention in disease-related molecular network. Mol Syst Biol 4:228, 2008.
Yeger-Lotem E, Margalit H. Detection of regulatory circuits by integrating the cellular networks of protein-protein interactions and transcription regulation. Nucleic Acids Res 31(20):6053-6061, 2003.
Young A, Whitehouse N, Cho J, Shaw C. Ontology traverser: an R package for GO analysis. Bioinformatics 21(2):275-276, 2005.
Yu H, Kim PM, Sprecher E, Trifonov V, Gerstein M. The importance of bottlenecks in protein networks: correlation with gene essentiality and expression dynamics. PLoS Comput Biol 3(4):e59, 2007.
Zhang C-C, Kast J. Applications of current proteomics techniques in modern drug design. Current Computer-Aided Drug Design 6(3):147-164, 2010.
Zhou Q, Kapoor M, Guo M, Belani R, Xu X, Kiosses WB, Hanan M, Park C, Armour E, Do M-H, Nangle LA, Schimmel P, Yang X-L. Orthogonal use of a human tRNA synthetase active site to achieve multifunctionality. Nat Struct Mol Biol 17(1):57-61, 2010.
[Discovery Medicine; ISSN: 1539-6509; Discov Med 11(57):133-143, February 2011.]