#### DMCA

## Statistical model checking in BioLab: applications to the automated analysis of T-Cell receptor signaling pathway (2008)

### Cached

### Download Links

- [www.cs.cmu.edu]
- [www.cs.cmu.edu]
- [www.ccbb.pitt.edu]
- [www.csb.pitt.edu]
- [www.ccbb.pitt.edu]
- [repository.cmu.edu]
- DBLP

### Other Repositories/Bibliography

Venue: | In CMSB’08 |

Citations: | 25 - 7 self |

### Citations

777 |
A general method for numerically simulating the stochastic time evolution of coupled chemical reactions
- Gillespie
- 1976
(Show Context)
Citation Context ...al logic. BioLab then statistically verifies the property using sequential hypothesis testing on executions sampled from the model. These samples are generated using variants of Gillespie’s algorithm =-=[15, 11, 31]-=-, which ensures that the executions are drawn from the “correct” underlying probability distribution. This, combined with the use of sequential hypothesis testing provides several guarantees. First, B... |

277 | Assembly of cell regulatory systems through protein interaction domains.
- Pawson, Nash
- 2003
(Show Context)
Citation Context ...cribe the transformations of molecules in the system possessing particular properties. The assumption underlying this modeling approach, which is consistent with the modularity of regulatory proteins =-=[27]-=-, is that interactions are governed by local context that can be captured in simple rules. Rules can be used to generate reaction networks that account comprehensively for the consequences of protein-... |

237 |
Sequential tests of statistical hypotheses
- Wald
- 1945
(Show Context)
Citation Context ...wo hypotheses K or H is accepted. 3.3 An Algorithmic Scheme Younes proposed a procedure to test H0 : p ≥ p0 against H1 : p ≤ p1 that is based on the sequential probability ratio test proposed by Wald =-=[30]-=-. The approach is briefly described below. Let Bi be a discrete random variable with a Bernoulli distribution. Such a variable can only take 2 values 0 and 1 with P r[Bi = 1] = p and P r[Bi = 0] =1 −... |

235 | Model-checking algorithms for continuous-time markov chains.
- Baier, Haverkort, et al.
- 2003
(Show Context)
Citation Context ...thm is numerical in the sense that it computes the exact probability for the system to satisfy φ and then compares it with the value of θ. Successful probabilistic model checking algorithms (see e.g. =-=[2, 14, 12, 8]-=-) and tools (see e.g. [26, 11]) have been proposed for various classes of systems, including (continuous time) Markov chains and Markov Decision Processes. The drawback with those approaches is that t... |

215 |
M.: The complexity of probabilistic verification
- Courcoubetis, Yannakakis
- 1995
(Show Context)
Citation Context ...thm is numerical in the sense that it computes the exact probability for the system to satisfy φ and then compares it with the value of θ. Successful probabilistic model checking algorithms (see e.g. =-=[2, 14, 12, 8]-=-) and tools (see e.g. [26, 11]) have been proposed for various classes of systems, including (continuous time) Markov chains and Markov Decision Processes. The drawback with those approaches is that t... |

105 |
Kinetic proofreading in T-cell receptor signal transduction
- McKeithan
- 1995
(Show Context)
Citation Context ...s has recently been developed by Lipniacki et al. [23], and serves as the basis for the experiments we conduct here using BioLab. This model extends previous simplified models of kinetic proofreading =-=[25]-=- and feedback regulation [28] by incorporating mechanistic detail about the involvement of specific signaling molecules. A schematic illustration of the model is presented in Fig. 5.2. Binding of pMHC... |

104 | Rule-based modelling of cellular signalling
- Danos, Feret, et al.
- 2007
(Show Context)
Citation Context ...tion of the language and underlying graph theory is provided in [4]. BioNetGen is similar to the κ-calculus, which has also been developed as a language for rule-based modeling of biochemical systems =-=[9]-=-. Other tools for rule-based modeling are reviewed in [18]. The syntax and semantics of BioNetGen have been thoroughly described in [13]. Briefly, a BioNetGen model is comprised of six basic elements ... |

87 |
BioNetGen: software for rule-based modeling of signal transduction based on the interactions of molecular domains
- Blinov, Faeder, et al.
(Show Context)
Citation Context ...ns. Examples of rule-based models of specific systems can be found in [16, 3, 1, 26]. BioNetGen is a software package that provides tools and a language for rule-based modeling of biochemical systems =-=[2, 13]-=-. A formal description of the language and underlying graph theory is provided in [4]. BioNetGen is similar to the κ-calculus, which has also been developed as a language for rule-based modeling of bi... |

85 | Validation of qualitative models of genetic regulatory networks by model checking: analysis of the nutritional stress response in E.Coli.
- Batt, Ropers, et al.
- 2005
(Show Context)
Citation Context ...red systems, and thus the majority of Model Checking algorithms are designed for such systems. Recently, however, there has been growing interest in the application of model checking to biology (e.g.,=-=[1, 4, 9, 10, 25, 27, 28, 34]-=-). Biological systems present new challenges in the context of formal verification. In particular, biological systems tend to give rise to highly parameterized models with stochastic dynamics. Biologi... |

77 | Rules for modeling signal-transduction systems.
- Hlavacek, Faeder, et al.
- 2006
(Show Context)
Citation Context ...rotein-protein interactions and other types of interactions that occur in biochemical systems can be modeled by formulating rules for each type of chemical transformation mediated by the interactions =-=[18]-=-. The rules can be viewed as definitions of reaction classes and used as generators of reactions, which describe the transformations of molecules in the system possessing particular properties. The as... |

75 | Prism 2.0: A tool for probabilistic model checking. In: QEST,
- Kwiatkowska, Norman, et al.
- 2004
(Show Context)
Citation Context ... an algorithm is numerical in the sense that it computes the exact probability for the system to satisfy φ and then compares it with the value of θ. Successful probabilistic model checking algorithms =-=[7, 20]-=-) have been proposed for various classes of systems, including (continuous time) Markov chains and Markov Decision Processes. The drawback with those approaches is that they compute the probability fo... |

66 | Symbolic Model Checking of Biochemical Networks. In:
- Chabrier, Fages
- 2003
(Show Context)
Citation Context ...red systems, and thus the majority of Model Checking algorithms are designed for such systems. Recently, however, there has been growing interest in the application of model checking to biology (e.g.,=-=[5, 6, 19, 21, 22]-=-). Biological systems present new challenges in the context of formal verification. In particular, biological systems tend to give rise to highly parameterized models with stochastic dynamics. Biologi... |

59 | Scalable simulation of cellular signalling networks.
- Danos, Feret, et al.
- 2007
(Show Context)
Citation Context ...y to maintain computational efficiency for large reaction networks, but is not practical for rule-based models that include oligomerization or attempt a comprehensive description of reaction networks =-=[18, 10]-=-. The NF method [10, 31] avoids explicit generation of species and reactions by simulating molecules as agents and has been shown to have per event cost that is independent of the number of possible s... |

59 | Model building and model checking for biochemical processes.
- Antoniotti, Policriti, et al.
- 2003
(Show Context)
Citation Context ...red systems, and thus the majority of Model Checking algorithms are designed for such systems. Recently, however, there has been growing interest in the application of model checking to biology (e.g.,=-=[1, 4, 9, 10, 25, 27, 28, 34]-=-). Biological systems present new challenges in the context of formal verification. In particular, biological systems tend to give rise to highly parameterized models with stochastic dynamics. Biologi... |

59 | Statistical model checking of black-box probabilistic systems.
- Sen, Viswanathan, et al.
- 2004
(Show Context)
Citation Context ...obabilistic model checking algorithms [7, 20]) have been proposed for various classes of systems, including (continuous time) Markov chains and Markov Decision Processes. The drawback with those approaches is that they compute the probability for all the executions of the system, which may not scale up for systems of large size. Another way to solve the probabilistic model checking problem is to use a statistical model checking algorithm. In the rest of this section, we recap the statistical model checking algorithmic scheme proposed by Younes in [33]. 3.2 Statistical Approach The approach in [33, 29] is based on hypothesis testing. The idea is to check the property φ on a sample set of simulations and to decide whether the system satisfies Pr≥θ(φ) based on the number of executions for which φ holds compared to the total number of executions in the sample set. With such an approach, we do not need to consider all the executions of the system. To determine whether M satisfies φ with a probability p ≥ θ, we can test the hypothesis H : p ≥ θ against K : p < θ. A test-based solution does not guarantee a correct result but it is possible to bound the probability of making an error. The strength... |

57 |
Liquor: A tool for qualitative and quantitative linear time analysis of reactive systems
- Ciesinski, Baier
- 2006
(Show Context)
Citation Context ...t computes the exact probability for the system to satisfy φ and then compares it with the value of θ. Successful probabilistic model checking algorithms (see e.g. [2, 14, 12, 8]) and tools (see e.g. =-=[26, 11]-=-) have been proposed for various classes of systems, including (continuous time) Markov chains and Markov Decision Processes. The drawback with those approaches is that they compute the probability fo... |

49 | Machine learning biochemical networks from temporal logic properties.
- Calzone, Chabrier-Rivier, et al.
- 2006
(Show Context)
Citation Context ...red systems, and thus the majority of Model Checking algorithms are designed for such systems. Recently, however, there has been growing interest in the application of model checking to biology (e.g.,=-=[5, 6, 19, 21, 22]-=-). Biological systems present new challenges in the context of formal verification. In particular, biological systems tend to give rise to highly parameterized models with stochastic dynamics. Biologi... |

47 | Checking Finite Traces using Alternating Automata
- Finkbeiner, Sipma
- 2001
(Show Context)
Citation Context ...imulator is used to generate stochastic traces and the trace verifier verifies each of them against the BLTL property. Our trace verifier is based on the translation from BLTL to alternating automata =-=[29, 14]-=-. The statistical model checker continues to simulate the BioNetGen model until a decision about the property has been made. Sequential Hypothesis Testing Algorithm BioNetgen Model BioNetgen Temporal ... |

46 | Analysis of signalling pathways using the PRISM model checker
- Calder, Vyshemirsky, et al.
(Show Context)
Citation Context ...red systems, and thus the majority of Model Checking algorithms are designed for such systems. Recently, however, there has been growing interest in the application of model checking to biology (e.g.,=-=[1, 4, 9, 10, 25, 27, 28, 34]-=-). Biological systems present new challenges in the context of formal verification. In particular, biological systems tend to give rise to highly parameterized models with stochastic dynamics. Biologi... |

43 | Hlavacek,W.S.: Rule-based modeling of biochemical systems with BioNetGen.
- Faeder, Blinov
- 2009
(Show Context)
Citation Context ...aper, we introduce a new tool, called BioLab, for formally reasoning about the behavior of stochastic dynamic models by integrating techniques from the field of Model Checking [13] into the BioNetGen =-=[18, 19]-=- ⋆ This research was sponsored by the Gigascale Systems Research Center (GSRC), the Semiconductor Research Corporation (SRC), the Office of Naval Research (ONR), the Naval Research Laboratory (NRL), t... |

40 |
A network model of early events in epidermal growth factor receptor signaling that accounts for combinatorial complexity
- Blinov, Faeder, et al.
- 2006
(Show Context)
Citation Context ...s. Rules can be used to generate reaction networks that account comprehensively for the consequences of protein-protein interactions. Examples of rule-based models of specific systems can be found in =-=[16, 3, 1, 26]-=-. BioNetGen is a software package that provides tools and a language for rule-based modeling of biochemical systems [2, 13]. A formal description of the language and underlying graph theory is provide... |

37 | On probabilistic computation tree logic
- Ciesinski, Größer
- 2004
(Show Context)
Citation Context ... an algorithm is numerical in the sense that it computes the exact probability for the system to satisfy φ and then compares it with the value of θ. Successful probabilistic model checking algorithms =-=[7, 20]-=-) have been proposed for various classes of systems, including (continuous time) Markov chains and Markov Decision Processes. The drawback with those approaches is that they compute the probability fo... |

37 |
W.S.: Mathematical and computational models of immune-receptor signaling.
- Goldstein, Faeder, et al.
- 2004
(Show Context)
Citation Context ...s. Rules can be used to generate reaction networks that account comprehensively for the consequences of protein-protein interactions. Examples of rule-based models of specific systems can be found in =-=[16, 3, 1, 26]-=-. BioNetGen is a software package that provides tools and a language for rule-based modeling of biochemical systems [2, 13]. A formal description of the language and underlying graph theory is provide... |

37 | Alternating automata and program verification.
- Vardi
- 1995
(Show Context)
Citation Context ...imulator is used to generate stochastic traces and the trace verifier verifies each of them against the BLTL property. Our trace verifier is based on the translation from BLTL to alternating automata =-=[29, 14]-=-. The statistical model checker continues to simulate the BioNetGen model until a decision about the property has been made. Sequential Hypothesis Testing Algorithm BioNetgen Model BioNetgen Temporal ... |

34 | Verification and Planning for Stochastic Processes with Asynchronous Events. PhD thesis,
- Younes
- 2005
(Show Context)
Citation Context ...he probabilistic model checking problem is to use a statistical model checking algorithm. In the rest of this section, we recap the statistical model checking algorithmic scheme proposed by Younes in =-=[32]-=-. 3.2 Statistical Approach The approach in [32] is based on hypothesis testing. The idea is to check the property φ on a sample set of simulations and to decide whether the system satisfies P r≥θ(φ) b... |

34 |
Algorithmic Algebraic Model Checking I: Challenges from Systems Biology.
- Piazza, Antoniotti, et al.
- 2005
(Show Context)
Citation Context |

30 | W.: Graph theory for rule-based modeling of biochemical networks
- Blinov, Yang, et al.
- 2005
(Show Context)
Citation Context ...NetGen is a software package that provides tools and a language for rule-based modeling of biochemical systems [2, 13]. A formal description of the language and underlying graph theory is provided in =-=[4]-=-. BioNetGen is similar to the κ-calculus, which has also been developed as a language for rule-based modeling of biochemical systems [9]. Other tools for rule-based modeling are reviewed in [18]. The ... |

30 |
E.: Simulation and verification for computational modelling of signalling pathways. In:
- Kwiatkowska, Norman, et al.
- 2006
(Show Context)
Citation Context ...red systems, and thus the majority of Model Checking algorithms are designed for such systems. Recently, however, there has been growing interest in the application of model checking to biology (e.g.,=-=[5, 6, 19, 21, 22]-=-). Biological systems present new challenges in the context of formal verification. In particular, biological systems tend to give rise to highly parameterized models with stochastic dynamics. Biologi... |

29 |
H.M.: Kinetic discrimination in T-cell activation.
- Rabinowitz, Beeson, et al.
- 1996
(Show Context)
Citation Context ...receptor interactions that are too short, positive feedback, which amplifies the response and makes it more switch-like, and negative feedback, which acts in concert with kinetic proofreading to dampen responses to weak stimulation and with positive feedback to enhance the stability of the inactive state. A computational model incorporating all three of these mechanisms has recently been developed by Lipniacki et al. [23], and serves as the basis for the experiments we conduct here using BioLab. This model extends previous simplified models of kinetic proofreading [25] and feedback regulation [28] by incorporating mechanistic detail about the involvement of specific signaling molecules. A schematic illustration of the model is presented in Fig. 5.2. Binding of pMHC to the TCR initiates a series of binding and phosphorylation events at the receptor that can lead either to activation or inhibition of the receptor depending on the strength of the stimulus, which is indicated along the kinetic proofreading axis. The rectangular box in the figure represents the TCR complex, which requires three components to make its passage to the activated form. These components are pMHC (P), doubly phosp... |

27 |
B.: The complexity of complexes in signal transduction.
- Hlavacek, Faeder, et al.
- 2003
(Show Context)
Citation Context ...Gen Proteins in cellular regulatory systems, because of their multicomponent composition, can interact in a combinatorial number of ways to generate myriad protein complexes, which are highly dynamic =-=[17]-=-. Protein-protein interactions and other types of interactions that occur in biochemical systems can be modeled by formulating rules for each type of chemical transformation mediated by the interactio... |

25 | W.S.: Graphical rule-based representation of signal-transduction networks. In:
- Faeder, Blinov, et al.
- 2005
(Show Context)
Citation Context ...paper, we introduce a new tool, called BioLab, for formally reasoning about the behavior of stochastic dynamic models by integrating techniques from the field of Model Checking [8] into the BioNetGen =-=[12, 13]-=- framework for rule-based modeling. We then use BioLab to verify the stochastic bistability of T-cell signalling. ⋆ This research was sponsored by the Gigascale Systems Research Center (GSRC), the Sem... |

22 | Rule-based modeling of biochemical networks
- Faeder, Blinov, et al.
- 2005
(Show Context)
Citation Context ...al logic. BioLab then statistically verifies the property using sequential hypothesis testing on executions sampled from the model. These samples are generated using variants of Gillespie’s algorithm =-=[15, 11, 31]-=-, which ensures that the executions are drawn from the “correct” underlying probability distribution. This, combined with the use of sequential hypothesis testing provides several guarantees. First, B... |

17 |
W.S.: Kinetic Monte Carlo method for rule-based modeling of biochemical networks
- Yang, Monine, et al.
- 2007
(Show Context)
Citation Context ...al logic. BioLab then statistically verifies the property using sequential hypothesis testing on executions sampled from the model. These samples are generated using variants of Gillespie’s algorithm =-=[15, 11, 31]-=-, which ensures that the executions are drawn from the “correct” underlying probability distribution. This, combined with the use of sequential hypothesis testing provides several guarantees. First, B... |

14 | Error control for probabilistic model checking
- Younes
- 2006
(Show Context)
Citation Context ...he probabilistic model checking problem is to use a statistical model checking algorithm. In the rest of this section, we recap the statistical model checking algorithmic scheme proposed by Younes in =-=[39, 40]-=-. 3.2 Statistical Approach The approach in [39, 40] is based on hypothesis testing. The idea is to check the property φ on a sample set of simulations and to decide whether the system satisfies Pr≥θ(φ... |

13 | S.K.: Symbolic approaches to finding control strategies in boolean networks. In:
- Langmead, Jha
- 2008
(Show Context)
Citation Context |

6 |
S.K.: Predicting protein folding kinetics via model checking. In:
- Langmead, Jha
- 2007
(Show Context)
Citation Context |

6 |
W.S.: Stochastic effects and bistability in T cell receptor signaling.
- Lipniacki, Hat, et al.
- 2008
(Show Context)
Citation Context ... (iii) We verify that a stochastic model of T-cell receptor signaling exhibits behaviors that are qualitatively different from those seen in an ordinary differential equation model of the same system =-=[23]-=-. In particular we verify that stochastic effects induce switching between two stable steady states of the system.2 BioNetGen Proteins in cellular regulatory systems, because of their multicomponent ... |

5 | J.M.: Structure-based kinetic models of modular signaling protein function: Focus on Shp2.
- Barua, Faeder, et al.
- 2007
(Show Context)
Citation Context ...s. Rules can be used to generate reaction networks that account comprehensively for the consequences of protein-protein interactions. Examples of rule-based models of specific systems can be found in =-=[16, 3, 1, 26]-=-. BioNetGen is a software package that provides tools and a language for rule-based modeling of biochemical systems [2, 13]. A formal description of the language and underlying graph theory is provide... |

3 |
Automatic generation of cellular networks with Moleculizer 1.0.
- Lok, Brent
- 2005
(Show Context)
Citation Context ...) [15] to sample the exact solution to the chemical master equations governing the species probabilities (GF-SSA). Both methods generate traces1 of the species concentrations as a function of time, but the GF-ODE algorithm is deterministic for a given initial state and set of system parameters, whereas each simulation run of GF-SSA from a given initial state represents a stochastic process and may generate a different trace. Like GF-SSA, OTF uses the Gillespie algorithm to generate traces but only generates species and reactions that are reachable within a small number of specified time steps [24, 11]. OTF was originally proposed as a way to maintain computational efficiency for large reaction networks, but is not practical for rule-based models that include oligomerization or attempt a comprehensive description of reaction networks [18, 10]. The NF method [10, 32] avoids explicit generation of species and reactions by simulating molecules as agents and has been shown to have per event cost that is independent of the number of possible species or reactions [10]. NF also relies on the SSA to sample reaction events that govern the evolution of the molecular agents. Because species are not ex... |

3 | Verifying omega-regular properties of markov chains
- Bustan, Rubin, et al.
- 2004
(Show Context)
Citation Context ...thm is numerical in the sense that it computes the exact probability for the system to satisfy φ and then compares it with the value of θ. Successful probabilistic model checking algorithms (see e.g. =-=[2, 14, 12, 8]-=-) and tools (see e.g. [26, 11]) have been proposed for various classes of systems, including (continuous time) Markov chains and Markov Decision Processes. The drawback with those approaches is that t... |

2 |
W.S.: Carbon fate maps for metabolic reactions.
- Mu, Williams, et al.
- 2007
(Show Context)
Citation Context |

2 |
W.S.: BioNetGen: software for rule-based modeling of signal transduction based on the interactions of molecular domains. Bioinformatics
- Blinov, Faeder, et al.
(Show Context)
Citation Context ...ons of molecules in the system possessing particular properties. The assumption underlying this modeling approach, which is consistent with the modularity of regulatory proteins [27], is that interactions are governed by local context that can be captured in simple rules. Rules can be used to generate reaction networks that account comprehensively for the consequences of protein-protein interactions. Examples of rule-based models of specific systems can be found in [16, 3, 1, 26]. BioNetGen is a software package that provides tools and a language for rule-based modeling of biochemical systems [2, 13]. A formal description of the language and underlying graph theory is provided in [4]. BioNetGen is similar to the κ-calculus, which has also been developed as a language for rule-based modeling of biochemical systems [9]. Other tools for rule-based modeling are reviewed in [18]. The syntax and semantics of BioNetGen have been thoroughly described in [13]. Briefly, a BioNetGen model is comprised of six basic elements that are defined in separate blocks in the input file: parameters, molecule types, seed species, reaction rules, observables, and actions. Molecules are the basic building blocks ... |

2 |
W.S.: Rule-based modeling of biochemical networks.
- Faeder, Blinov, et al.
- 2005
(Show Context)
Citation Context ...and parameter values. Model checking algorithms targeting biological applications must therefore apply to stochastic, multi-parameter models. BioLab models stochastic biochemical systems using the BioNetGen modeling language. The set of initial states (i.e., S0) comprise a user-specified set of initial conditions and parameter values. Properties are expressed in probabilistic bounded linear temporal logic. BioLab then statistically verifies the property using sequential hypothesis testing on executions sampled from the model. These samples are generated using variants of Gillespie’s algorithm [15, 11, 32], which ensures that the executions are drawn from the “correct” underlying probability distribution. This, combined with the use of sequential hypothesis testing provides several guarantees. First, BioLab can bound the probability of Type-I (i.e., false-positive) and Type-II (i.e., false-negative) errors, with regard to the predictions it makes. These error bounds are specified by the user. Second, BioLab is optimal in the sense that it generates the minimum number of executions necessary to determine whether a given property is satisfied. The number of required executions varies depending on... |