Rule-based modelling of biological systems using regulated rewriting

In systems biology, models play a crucial role in understanding studied systems. There are many modelling approaches, among which rewriting systems provide a framework for describing systems on a mechanistic level. Describing biochemical processes often requires incorporating knowledge on an abstract level to simplify the system description or substitute the missing details. For this purpose, we present regulation mechanisms, an extension of this formalism with additional controls on the rewriting process. We introduce several regulation mechanisms and apply them to a rule-based language, a notation suitable for modelling biological phenomena. Finally, we demonstrate the usage of such regulations on several case studies from the biochemical domain.

Comparison of rule- and ordinary differential equation-based dynamic model of DARPP-32 signalling network

Dynamic modelling has considerably improved our understanding of complex molecular mechanisms. Ordinary differential equations (ODEs) are the most detailed and popular approach to modelling the dynamics of molecular systems. However, their application in signalling networks, characterised by multi-state molecular complexes, can be prohibitive. Contemporary modelling methods, such as rule- based (RB) modelling, have addressed these issues. The advantages of RB modelling over ODEs have been presented and discussed in numerous reviews. In this study, we conduct a direct comparison of the time courses of a molecular system founded on the same reaction network but encoded in the two frameworks. To make such a comparison, a set of reactions that underlie an ODE model was manually encoded in the Kappa language, one of the RB implementations. A comparison of the models was performed at the level of model specification and dynamics, acquired through model simulations. In line with previous reports, we confirm that the Kappa model recapitulates the general dynamics of its ODE counterpart with minor differences. These occur when molecules have multiple sites binding the same interactor. Furthermore, activation of these molecules in the RB model is slower than in the ODE one. As reported for other molecular systems, we find that, also for the DARPP-32 reaction network, the RB representation offers a more expressive and flexible syntax that facilitates access to fine details of the model, easing model reuse. In parallel with these analyses, we report a refactored model of the DARPP-32 interaction network that can serve as a canvas for the development of more complex dynamic models to study this important molecular system.

Rule-based epidemic models

Rule-based models generalise reaction-based models with reagents that have internal state and may be bound together to form complexes, as in chemistry. An important class of system that would be intractable if expressed as reactions or ordinary differential equations can be efficiently simulated when expressed as rules. In this paper we demonstrate the utility of the rule-based approach for epidemiological modelling presenting a suite of seven models illustrating the spread of infectious disease under different scenarios: wearing masks, infection via fomites and prevention by hand-washing, the concept of vector-borne diseases, testing and contact tracing interventions, disease propagation within motif-structured populations with shared environments such as schools, and superspreading events. Rule-based models allow to combine transparent modelling approach with scalability and compositionality and therefore can facilitate the study of aspects of infectious disease propagation in a richer context than would otherwise be feasible.

Analysis of Fn14-NF-κB signaling response dynamics using a mechanistic model

Fn14 is a transmembrane receptor protein belonging to the tumor necrosis factor receptor (TNFR) superfamily. Many experimental reports have shown that crosslinking of the receptor by its extracellular ligand TWEAK induces prolonged activation of transcription factor NF-κB. This behavior is distinct from TNF-α receptor, which is a more well-characterized member of the TNFR family. TNF-α receptor, despite sharing many similar molecular interactions with Fn14, only transiently activates NF-κB in response to TNF-α stimulation. Here, we investigate molecular mechanisms that enable Fn14 to display such distinctive behavior. In particular, we focus on two specific features of the Fn14 pathway that potentially give rise to a positive feedback regulation and differentiate it from the TNF-α receptor signaling. By developing a mechanistic model, we analyze how these features may determine the dynamics of an Fn14-NF-κB response. Our analysis reveals that stimulation of Fn14 by TWEAK may generate highly non-linear dynamics, including stable limit cycles and bistable responses. The type of response depends both on the strength and duration of a TWEAK signal. Our predictions and analyses also show that the molecular interactions underlying the positive feedback explain the prolonged activation of NF-κB under certain parameter regimes. In light of the model predictions, we propose possible deregulations of Fn14 leading to its overexpression in solid tumors and tissue injuries.

Markov Chain Aggregation and Its Application to Rule-Based Modelling

Rule-based modelling allows to represent molecular interactions in a compact and natural way. The underlying molecular dynamics, by the laws of stochastic chemical kinetics, behaves as a continuous-time Markov chain. However, this Markov chain enumerates all possible reaction mixtures, rendering the analysis of the chain computationally demanding and often prohibitive in practice. We here describe how it is possible to efficiently find a smaller, aggregate chain, which preserves certain properties of the original one. Formal methods and lumpability notions are used to define algorithms for automated and efficient construction of such smaller chains (without ever constructing the original ones). We here illustrate the method on an example and we discuss the applicability of the method in the context of modelling large signaling pathways.

Strategy-Driven Exploration for Rule-Based Models of Biochemical Systems with PORGY

This chapter presents PORGY-an interactive visual environment for rule-based modelling of biochemical systems. We model molecules and molecule interactions as port graphs and port graph rewrite rules, respectively. We use rewriting strategies to control which rules to apply, and where and when to apply them. Our main contributions to rule-based modelling of biochemical systems lie in the strategy language and the associated visual and interactive features offered by PORGY. These features facilitate an exploratory approach to test different ways of applying the rules while recording the model evolution, and tracking and plotting parameters. We illustrate PORGY's features with a study of the role of a scaffold protein in RAF/MEK/ERK signalling.

Evaluation of a rule-based method for epidemiological document classification towards the automation of systematic reviews

INTRODUCTION

Most data extraction efforts in epidemiology are focused on obtaining targeted information from clinical trials. In contrast, limited research has been conducted on the identification of information from observational studies, a major source for human evidence in many fields, including environmental health. The recognition of key epidemiological information (e.g., exposures) through text mining techniques can assist in the automation of systematic reviews and other evidence summaries.

METHOD

We designed and applied a knowledge-driven, rule-based approach to identify targeted information (study design, participant population, exposure, outcome, confounding factors, and the country where the study was conducted) from abstracts of epidemiological studies included in several systematic reviews of environmental health exposures. The rules were based on common syntactical patterns observed in text and are thus not specific to any systematic review. To validate the general applicability of our approach, we compared the data extracted using our approach versus hand curation for 35 epidemiological study abstracts manually selected for inclusion in two systematic reviews.

RESULTS

The returned F-score, precision, and recall ranged from 70% to 98%, 81% to 100%, and 54% to 97%, respectively. The highest precision was observed for exposure, outcome and population (100%) while recall was best for exposure and study design with 97% and 89%, respectively. The lowest recall was observed for the population (54%), which also had the lowest F-score (70%).

CONCLUSION

The generated performance of our text-mining approach demonstrated encouraging results for the identification of targeted information from observational epidemiological study abstracts related to environmental exposures. We have demonstrated that rules based on generic syntactic patterns in one corpus can be applied to other observational study design by simple interchanging the dictionaries aiming to identify certain characteristics (i.e., outcomes, exposures). At the document level, the recognised information can assist in the selection and categorization of studies included in a systematic review.

A rule-based model of insulin signalling pathway

BACKGROUND

The insulin signalling pathway (ISP) is an important biochemical pathway, which regulates some fundamental biological functions such as glucose and lipid metabolism, protein synthesis, cell proliferation, cell differentiation and apoptosis. In the last years, different mathematical models based on ordinary differential equations have been proposed in the literature to describe specific features of the ISP, thus providing a description of the behaviour of the system and its emerging properties. However, protein-protein interactions potentially generate a multiplicity of distinct chemical species, an issue referred to as "combinatorial complexity", which results in defining a high number of state variables equal to the number of possible protein modifications. This often leads to complex, error prone and difficult to handle model definitions.

RESULTS

In this work, we present a comprehensive model of the ISP, which integrates three models previously available in the literature by using the rule-based modelling (RBM) approach. RBM allows for a simple description of a number of signalling pathway characteristics, such as the phosphorylation of signalling proteins at multiple sites with different effects, the simultaneous interaction of many molecules of the signalling pathways with several binding partners, and the information about subcellular localization where reactions take place. Thanks to its modularity, it also allows an easy integration of different pathways. After RBM specification, we simulated the dynamic behaviour of the ISP model and validated it using experimental data. We the examined the predicted profiles of all the active species and clustered them in four clusters according to their dynamic behaviour. Finally, we used parametric sensitivity analysis to show the role of negative feedback loops in controlling the robustness of the system.

CONCLUSIONS

The presented ISP model is a powerful tool for data simulation and can be used in combination with experimental approaches to guide the experimental design. The model is available at http://sysbiobig.dei.unipd.it/ was submitted to Biomodels Database ( https://www.ebi.ac.uk/biomodels-main/ # MODEL 1604100005).

Using local lexicalized rules to identify heart disease risk factors in clinical notes

Heart disease is the leading cause of death globally and a significant part of the human population lives with it. A number of risk factors have been recognized as contributing to the disease, including obesity, coronary artery disease (CAD), hypertension, hyperlipidemia, diabetes, smoking, and family history of premature CAD. This paper describes and evaluates a methodology to extract mentions of such risk factors from diabetic clinical notes, which was a task of the i2b2/UTHealth 2014 Challenge in Natural Language Processing for Clinical Data. The methodology is knowledge-driven and the system implements local lexicalized rules (based on syntactical patterns observed in notes) combined with manually constructed dictionaries that characterize the domain. A part of the task was also to detect the time interval in which the risk factors were present in a patient. The system was applied to an evaluation set of 514 unseen notes and achieved a micro-average F-score of 88% (with 86% precision and 90% recall). While the identification of CAD family history, medication and some of the related disease factors (e.g. hypertension, diabetes, hyperlipidemia) showed quite good results, the identification of CAD-specific indicators proved to be more challenging (F-score of 74%). Overall, the results are encouraging and suggested that automated text mining methods can be used to process clinical notes to identify risk factors and monitor progression of heart disease on a large-scale, providing necessary data for clinical and epidemiological studies.