Morgan Magnin

Under-approximation of Reachability in Multivalued Asynchronous Networks

The Process Hitting is a recently introduced framework designed for the modelling of concurrent systems. Its originality lies in a compact representation of both components of the model and its corresponding actions: each action can modify the status of a component, and is conditioned by the status of at most one other component. This allowed to define very efficient static analysis based on local causality to compute reachability properties. However, in the case of cooperations between components (for example, when two components are supposed to interact with a third one only when they are in a given configuration), the approach leads to an over-approximated interleaving between actions, because of the pure asynchronous semantics of the model. To address this issue, we propose an extended definition of the framework, including priority classes for actions. In this paper, we focus on a restriction of the Process Hitting with two classes of priorities and a specific behaviour of the components, that is sufficient to tackle the aforementioned problem of cooperations. We show that this class of Process Hitting models allows to represent any Asynchronous Discrete Networks, either Boolean or multivalued. Then we develop a new refinement for the under-approximation of the static analysis to give accurate results for this class of Process Hitting models. Our method thus allows to efficiently under-approximate reachability properties in Asynchronous Discrete Networks; it is in particular conclusive on reachability properties in a 94 components Boolean network, which is unprecedented.

Tuning Temporal Features within the Stochastic π-Calculus

The stochastic π-calculus is a formalism that has been used for modeling complex dynamical systems where the stochasticity and the delay of transitions are important features, such as in the case of biochemical reactions. Commonly, durations of transitions within stochastic π-calculus models follow an exponential law. The underlying dynamics of such models are expressed in terms of continuous-time Markov chains, which can then be efficiently simulated and model-checked. However, the exponential law comes with a huge variance, making it difficult to model systems with accurate temporal constraints. In this paper, a technique for tuning temporal features within the stochastic π-calculus is presented. This method relies on the introduction of a stochasticity absorption factor by replacing the exponential distribution with the Erlang distribution, which is a sum of exponential random variables. This paper presents a construction of the stochasticity absorption factor in the classical stochastic π-calculus with exponential rates. Tools for manipulating the stochasticity absorption factor and its link with timed intervals for firing transitions are also presented. Finally, the model-checking of such designed models is tackled by supporting the stochasticity absorption factor in a translation from the stochastic π-calculus to the probabilistic model checker PRISM.

Abstract Interpretation of Dynamics of Biological Regulatory Networks

Analysing dynamics of large biological regulatory networks (BRNs) calls for innovative methods to cope with the state space explosion. Static analysis and abstract interpretation techniques seem promising approaches. In this paper, we address the Process Hitting framework, that has been shown of interest to model dynamics of BRNs with discrete values. We propose to take profit from the particular structures of Process Hitting to build efficient static analyses. We introduce a novel and original method to decide the reachability of the state of a component within a BRN modelled in Process Hitting. The decision is achieved by abstract interpretation and static analysis of Process Hittings. The scalability of our approach is illustrated by its application to the analysis of a BRN with 40 components.

Learning delayed influences of biological systems.

Boolean networks are widely used model to represent gene interactions and global dynamical behavior of gene regulatory networks. To understand the memory effect involved in some interactions between biological components, it is necessary to include delayed influences in the model. In this paper, we present a logical method to learn such models from sequences of gene expression data. This method analyzes each sequence one by one to iteratively construct a Boolean network that captures the dynamics of these observations. To illustrate the merits of this approach, we apply it to learning real data from bioinformatic literature. Using data from the yeast cell cycle, we give experimental results and show the scalability of the method. We show empirically that using this method we can handle millions of observations and successfully capture delayed influences of Boolean networks.

Exhaustive analysis of dynamical properties of Biological Regulatory Networks with Answer Set Programming

The combination of numerous simple influences between the components of a Biological Regulatory Network (BRN) often leads to behaviors that cannot be grasped intuitively. They thus call for the development of proper mathematical methods to delineate their dynamical properties. As a consequence, formal methods and computer tools for the modeling and simulation of BRNs become essential. Our recently introduced discrete formalism called the Process Hitting (PH), a restriction of synchronous automata networks, is notably suitable to such study. In this paper, we propose a new logical approach to perform model-checking of dynamical properties of BRNs modeled in PH. Our work here focuses on state reachability properties on the one hand, and on the identification of fixed points on the other hand. The originality of our model-checking approach relies in the exhaustive enumeration of all possible simulations verifying the dynamical properties thanks to the use of Answer Set Programming.

Sufficient conditions for reachability in automata networks with priorities

In this paper, we develop a framework for an efficient under-approximation of the dynamics of Asynchronous Automata Networks (AANs). An AAN is an Automata Network with synchronised transitions between automata, where each transition changes the local state of exactly one automaton (but any number of synchronising local states are allowed). The work we propose here is based on static analysis by abstract interpretation, which allows to prove that reaching a state with a given property is possible, without the same computational cost of usual model checkers: the complexity is polynomial with the total number of local states and exponential with the number of local states within a single automaton. Furthermore, we address AANs with classes of priorities, and give an encoding into AANs without priorities, thus extending the application range of our under-approximation. Finally, we illustrate our method for the model checking of large-scale biological networks.

Formalization of resilience for constraint-based dynamic systems

Many researchers in different fields are interested in building resilient systems that can absorb shocks and recover from damages caused by unexpected large-scale events. Existing approaches mainly focus on the evaluation of the resilience of systems from a qualitative point of view, or pay particular attention to some domain-dependent aspects of the resilience. In this paper, we introduce a very general, abstract computational model rich enough to represent a large class of constraint-based dynamic systems. Taking our inspiration from the literature, we propose a simple parameterized property which captures the main features of resilience independently from a particular application domain, and we show how to assess the resilience of a constraint-based dynamic system through this new resilience property.

Linking Discrete and Stochastic Models: The Chemical Master Equation as a Bridge between Process Hitting and Proper Generalized Decomposition

Modeling frameworks bring structure and analysis tools to large and non-intuitive systems but come with certain inherent assumptions and limitations, sometimes to an inhibitive extent. By building bridges in existing models, we can exploit the advantages of each, widening the range of analysis possible for larger, more detailed models of gene regulatory networks. In this paper, we create just such a link between Process Hitting [6,7,8], a recently introduced discrete framework, and the Chemical Master Equation in such a way that allows the application of powerful numerical techniques, namely Proper Generalized Decomposition [1,2,3], to overcome the curse of dimensionality. With these tools in hand, one can exploit the formal analysis of discrete models without sacrificing the ability to obtain a full space state solution, widening the scope of analysis and interpretation possible. As a demonstration of the utility of this methodology, we have applied it here to the p53-mdm2 network [4,5], a widely studied biological regulatory network.

Learning Multi-valued Biological Models with Delayed Influence from Time-Series Observations

Delayed effects are important in modeling biological systems, and timed Boolean networks have been proposed for such a framework. Yet it is not an easy task to design such Boolean models with delays precisely. Recently, an attempt to learn timed Boolean networks has been made in Ribeiro et al 2015 in the framework of learning state transition rules from time-series data. However, this approach still has two limitations: (1) The maximum delay has to be given as input to the algorithm, (2) The possible value of each state is assumed to be Boolean, i.e., twovalued. In this paper, we extend the previous learning mechanism to overcome these limitations. We propose an algorithm to learn multi-valued biological models with delayed influence by automatically tuning the delay. The delay is determined so as to minimally explain the necessary influences. The merits of our approach is then verified on benchmarks coming from the DREAM4 challenge.

ASP-based method for the enumeration of attractors in non-deterministic synchronous and asynchronous multi-valued networks

BackgroundThis paper addresses the problem of finding attractors in biological regulatory networks. We focus here on non-deterministic synchronous and asynchronous multi-valued networks, modeled using automata networks (AN). AN is a general and well-suited formalism to study complex interactions between different components (genes, proteins,...). An attractor is a minimal trap domain, that is, a part of the state-transition graph that cannot be escaped. Such structures are terminal components of the dynamics and take the form of steady states (singleton) or complex compositions of cycles (non-singleton). Studying the effect of a disease or a mutation on an organism requires finding the attractors in the model to understand the long-term behaviors.ResultsWe present a computational logical method based on answer set programming (ASP) to identify all attractors. Performed without any network reduction, the method can be applied on any dynamical semantics. In this paper, we present the two most widespread non-deterministic semantics: the asynchronous and the synchronous updating modes. The logical approach goes through a complete enumeration of the states of the network in order to find the attractors without the necessity to construct the whole state-transition graph. We realize extensive computational experiments which show good performance and fit the expected theoretical results in the literature.ConclusionThe originality of our approach lies on the exhaustive enumeration of all possible (sets of) states verifying the properties of an attractor thanks to the use of ASP. Our method is applied to non-deterministic semantics in two different schemes (asynchronous and synchronous). The merits of our methods are illustrated by applying them to biological examples of various sizes and comparing the results with some existing approaches. It turns out that our approach succeeds to exhaustively enumerate on a desktop computer, in a large model (100 components), all existing attractors up to a given size (20 states). This size is only limited by memory and computation time.