Contemporary computational approaches to equation discovery enable automated design of dynamic systems and thus extend their application scope towards the field of synthetic biology. It deploys engineering approaches to manage and build new, artificial biological systems. The contribution presents a new approach to design or automatic construction of in-silico models, the properties of which are in accordance with predefined user requirements. The approach combines process-based modeling and multi-criteria optimization. First, it makes it easy to specify a plethora of alternative system designs using the library of components and a template plan, while multi-criteria optimization is used to select the most appropriate design by taking into account multiple criteria based on the required model properties. The new approach is shown to be extremely useful for designing deterministic and stochastic systems with different required properties (e.g., oscillatory behavior). It has been successfully used for reconstructing known designs from the related literature and proposing new (so far unknown) designs. The results indicate that the approach is applicable to relevant design problems, such as planning an in-silico model of targeted and timely delivery of drugs to cells.
COBISS.SI-ID: 29806119
The paper reports the results of an extensive empirical analysis of the performance of 114 variants of the generalized relevance network approach on 47 tasks of gene network inference from time-series data and 39 tasks of gene network inference from steady-state data. The generalized relevance network approach to network inference reconstructs network links based on the strength of associations between data in individual network nodes. It can reconstruct undirected networks, i.e., relevance networks, sensu stricto, as well as directed networks, referred to as causal relevance networks. The generalized approach allows the use of an arbitrary measure of pairwise association between nodes, an arbitrary scoring scheme that transforms the associations into weights of the network links, and a method for inferring the directions of the links. While this makes the approach powerful and flexible, it introduces the challenge of finding a combination of components that would perform well on a given inference task. The results of our empirical analysis show that the association measures based on correlation, combined with a particular scoring scheme of asymmetric weighting, lead to optimal performance of the relevance network approach in the general case. In the two special cases of inference tasks involving short time-series data and/or large networks, association measures based on identifying qualitative trends in the time series are more appropriate.
COBISS.SI-ID: 31698727
We have developed a new network reconstruction approach that relies on decoupling approximation of network dynamics. Decoupling approximation consists of modeling the dynamics for each node individually, rather than for the entire network at once. We model the node dynamics using equation discovery methods for automated modeling of dynamical systems that minimize the discrepancy between the simulated and the observed trajectories. An important aspect of the proposed method is that it makes no assumptions about the properties of the trajectories at hand. The experimental evaluation of the method provides strong evidence that its performance is largely independent of the dynamical regime at hand. Of crucial relevance for practical applications, we also find our method to be extremely robust to both length and resolution of the trajectories and relatively insensitive to noise.
COBISS.SI-ID: 31814439
We have designed a novel method for network reconstruction based on feature ranking. We assign a feature to each network node based on the values of the trajectories (time-series) of the observed node dynamics. In turn, for a selected feature, we rank the features of all the other nodes with respect to their relevance for predicting the feature of the selected node. In this way, we extract information on what other nodes are most likely to be connected with the given one. The key property of the method is that it requires no assumption on the knowledge of interaction functions or the dynamical model of the network, and that it makes no hypotheses on the nature of the available trajectories (data). We test the performance of our method using networks of coupled logistic maps and obtain good results for a range of coupling strengths and network sizes. In addition, our method is able to perform well even for relatively short trajectories and it is fairly robust to noise.
COBISS.SI-ID: 32629031
We address the task of selecting an optimal model structure from a user-specified finite set of alternative, candidate model structures. We formulate the structure selection task as a combinatorial search problem. We propose a mapping of the set of candidate model structures to a fixed-length, vector representation allowing the use of an arbitrary search algorithm as a solver of the structure selection task. We perform a comparative analysis of the performance of thirteen variants of several search algorithms, ranging from ones with high intensification, i.e., focus on neighborhood of the best candidate solutions, to ones with high diversification, i.e., focus on covering the entire search space. The empirical analysis involves eight tasks of reconstructing known models of dynamical systems from synthetic and measured data. The results of the analysis show that search algorithms involving moderate diversification methods have superior performance on the structure selection task. The empirical analysis also reveals that this finding is related to specific properties of the search space of candidate model structures.
COBISS.SI-ID: 32975911