TY - JOUR
AB - We introduce propagation models (PMs), a formalism able to express several kinds of equations that describe the behavior of biochemical reaction networks. Furthermore, we introduce the propagation abstract data type (PADT), which separates concerns regarding different numerical algorithms for the transient analysis of biochemical reaction networks from concerns regarding their implementation, thus allowing for portable and efficient solutions. The state of a propagation abstract data type is given by a vector that assigns mass values to a set of nodes, and its (next) operator propagates mass values through this set of nodes. We propose an approximate implementation of the (next) operator, based on threshold abstraction, which propagates only "significant" mass values and thus achieves a compromise between efficiency and accuracy. Finally, we give three use cases for propagation models: the chemical master equation (CME), the reaction rate equation (RRE), and a hybrid method that combines these two equations. These three applications use propagation models in order to propagate probabilities and/or expected values and variances of the model's variables.
AU - Henzinger, Thomas A
AU - Mateescu, Maria
ID - 2302
IS - 2
JF - IEEE ACM Transactions on Computational Biology and Bioinformatics
TI - The propagation approach for computing biochemical reaction networks
VL - 10
ER -
TY - CONF
AB - Continuous-time Markov chains (CTMC) with their rich theory and efficient simulation algorithms have been successfully used in modeling stochastic processes in diverse areas such as computer science, physics, and biology. However, systems that comprise non-instantaneous events cannot be accurately and efficiently modeled with CTMCs. In this paper we define delayed CTMCs, an extension of CTMCs that allows for the specification of a lower bound on the time interval between an event's initiation and its completion, and we propose an algorithm for the computation of their behavior. Our algorithm effectively decomposes the computation into two stages: a pure CTMC governs event initiations while a deterministic process guarantees lower bounds on event completion times. Furthermore, from the nature of delayed CTMCs, we obtain a parallelized version of our algorithm. We use our formalism to model genetic regulatory circuits (biological systems where delayed events are common) and report on the results of our numerical algorithm as run on a cluster. We compare performance and accuracy of our results with results obtained by using pure CTMCs. © 2012 Springer-Verlag.
AU - Guet, Calin C
AU - Gupta, Ashutosh
AU - Henzinger, Thomas A
AU - Mateescu, Maria
AU - Sezgin, Ali
ID - 3136
TI - Delayed continuous time Markov chains for genetic regulatory circuits
VL - 7358
ER -
TY - JOUR
AB - Background
The chemical master equation (CME) is a system of ordinary differential equations that describes the evolution of a network of chemical reactions as a stochastic process. Its solution yields the probability density vector of the system at each point in time. Solving the CME numerically is in many cases computationally expensive or even infeasible as the number of reachable states can be very large or infinite. We introduce the sliding window method, which computes an approximate solution of the CME by performing a sequence of local analysis steps. In each step, only a manageable subset of states is considered, representing a "window" into the state space. In subsequent steps, the window follows the direction in which the probability mass moves, until the time period of interest has elapsed. We construct the window based on a deterministic approximation of the future behavior of the system by estimating upper and lower bounds on the populations of the chemical species.
Results
In order to show the effectiveness of our approach, we apply it to several examples previously described in the literature. The experimental results show that the proposed method speeds up the analysis considerably, compared to a global analysis, while still providing high accuracy.
Conclusions
The sliding window method is a novel approach to address the performance problems of numerical algorithms for the solution of the chemical master equation. The method efficiently approximates the probability distributions at the time points of interest for a variety of chemically reacting systems, including systems for which no upper bound on the population sizes of the chemical species is known a priori.
AU - Wolf, Verena
AU - Goel, Rushil
AU - Mateescu, Maria
AU - Henzinger, Thomas A
ID - 3834
IS - 42
JF - BMC Systems Biology
TI - Solving the chemical master equation using sliding windows
VL - 4
ER -
TY - CONF
AB - Within systems biology there is an increasing interest in the stochastic behavior of biochemical reaction networks. An appropriate stochastic description is provided by the chemical master equation, which represents a continuous- time Markov chain (CTMC).
Standard Uniformization (SU) is an efficient method for the transient analysis of CTMCs. For systems with very different time scales, such as biochemical reaction networks, SU is computationally expensive. In these cases, a variant of SU, called adaptive uniformization (AU), is known to reduce the large number of iterations needed by SU. The additional difficulty of AU is that it requires the solution of a birth process.
In this paper we present an on-the-fly variant of AU, where we improve the original algorithm for AU at the cost of a small approximation error. By means of several examples, we show that our approach is particularly well-suited for biochemical reaction networks.
AU - Didier, Frédéric
AU - Henzinger, Thomas A
AU - Mateescu, Maria
AU - Wolf, Verena
ID - 3843
IS - 6
TI - Fast adaptive uniformization of the chemical master equation
VL - 4
ER -
TY - CONF
AB - We present an on-the-fly abstraction technique for infinite-state continuous -time Markov chains. We consider Markov chains that are specified by a finite set of transition classes. Such models naturally represent biochemical reactions and therefore play an important role in the stochastic modeling of biological systems. We approximate the transient probability distributions at various time instances by solving a sequence of dynamically constructed abstract models, each depending on the previous one. Each abstract model is a finite Markov chain that represents the behavior of the original, infinite chain during a specific time interval. Our approach provides complete information about probability distributions, not just about individual parameters like the mean. The error of each abstraction can be computed, and the precision of the abstraction refined when desired. We implemented the algorithm and demonstrate its usefulness and efficiency on several case studies from systems biology.
AU - Thomas Henzinger
AU - Maria Mateescu
AU - Wolf, Verena
ID - 4453
TI - Sliding-window abstraction for infinite Markov chains
VL - 5643
ER -
TY - CONF
AB - Molecular noise, which arises from the randomness of the discrete events in the cell, significantly influences fundamental biological processes. Discrete -state continuous-time stochastic models (CTMC) can be used to describe such effects, but the calculation of the probabilities of certain events is computationally expensive.
We present a comparison of two analysis approaches for CTMC. On one hand, we estimate the probabilities of interest using repeated Gillespie simulation and determine the statistical accuracy that we obtain. On the other hand, we apply a numerical reachability analysis that approximates the probability distributions of the system at several time instances. We use examples of cellular processes to demonstrate the superiority of the reachability analysis if accurate results are required.
AU - Didier, Frédéric
AU - Thomas Henzinger
AU - Maria Mateescu
AU - Wolf, Verena
ID - 4535
TI - Approximation of event probabilities in noisy cellular processes
VL - 5688
ER -
TY - CONF
AB - We introduce bounded asynchrony, a notion of concurrency tailored to the modeling of biological cell-cell interactions. Bounded asynchrony is the result of a scheduler that bounds the number of steps that one process gets ahead of other processes; this allows the components of a system to move independently while keeping them coupled. Bounded asynchrony accurately reproduces the experimental observations made about certain cell-cell interactions: its constrained nondeterminism captures the variability observed in cells that, although equally potent, assume distinct fates. Real-life cells are not “scheduled”, but we show that distributed real-time behavior can lead to component interactions that are observationally equivalent to bounded asynchrony; this provides a possible mechanistic explanation for the phenomena observed during cell fate specification.
We use model checking to determine cell fates. The nondeterminism of bounded asynchrony causes state explosion during model checking, but partial-order methods are not directly applicable. We present a new algorithm that reduces the number of states that need to be explored: our optimization takes advantage of the bounded-asynchronous progress and the spatially local interactions of components that model cells. We compare our own communication-based reduction with partial-order reduction (on a restricted form of bounded asynchrony) and experiments illustrate that our algorithm leads to significant savings.
AU - Fisher, Jasmin
AU - Thomas Henzinger
AU - Maria Mateescu
AU - Piterman, Nir
ID - 4527
TI - Bounded asynchrony: Concurrency for modeling cell-cell interactions
VL - 5054
ER -