RULEMINER: A KNOWLEDGE SYSTEM FOR SUPPORTING HIGH-THROUGHPUT PROTEIN FUNCTION ANNOTATIONS

Mathematical and Computational Modeling in Complex Biological Systems

The biological process and molecular functions involved in the cancer progression remain difficult to understand for biologists and clinical doctors. Recent developments in high-throughput technologies urge the systems biology to achieve more precise models for complex diseases. Computational and mathematical models are gradually being used to help us understand the omics data produced by high-throughput experimental techniques. The use of computational models in systems biology allows us to explore the pathogenesis of complex diseases, improve our understanding of the latent molecular mechanisms, and promote treatment strategy optimization and new drug discovery. Currently, it is urgent to bridge the gap between the developments of high-throughput technologies and systemic modeling of the biological process in cancer research. In this review, we firstly studied several typical mathematical modeling approaches of biological systems in different scales and deeply analyzed their characteristics, advantages, applications, and limitations. Next, three potential research directions in systems modeling were summarized. To conclude, this review provides an update of important solutions using computational modeling approaches in systems biology.

SBML qualitative models: a model representation format and infrastructure to foster interactions between qualitative modelling formalisms and tools

BACKGROUND

Qualitative frameworks, especially those based on the logical discrete formalism, are increasingly used to model regulatory and signalling networks. A major advantage of these frameworks is that they do not require precise quantitative data, and that they are well-suited for studies of large networks. While numerous groups have developed specific computational tools that provide original methods to analyse qualitative models, a standard format to exchange qualitative models has been missing.

RESULTS

We present the Systems Biology Markup Language (SBML) Qualitative Models Package ("qual"), an extension of the SBML Level 3 standard designed for computer representation of qualitative models of biological networks. We demonstrate the interoperability of models via SBML qual through the analysis of a specific signalling network by three independent software tools. Furthermore, the collective effort to define the SBML qual format paved the way for the development of LogicalModel, an open-source model library, which will facilitate the adoption of the format as well as the collaborative development of algorithms to analyse qualitative models.

CONCLUSIONS

SBML qual allows the exchange of qualitative models among a number of complementary software tools. SBML qual has the potential to promote collaborative work on the development of novel computational approaches, as well as on the specification and the analysis of comprehensive qualitative models of regulatory and signalling networks.

Modeling approaches for qualitative and semi-quantitative analysis of cellular signaling networks

A central goal of systems biology is the construction of predictive models of bio-molecular networks. Cellular networks of moderate size have been modeled successfully in a quantitative way based on differential equations. However, in large-scale networks, knowledge of mechanistic details and kinetic parameters is often too limited to allow for the set-up of predictive quantitative models.Here, we review methodologies for qualitative and semi-quantitative modeling of cellular signal transduction networks. In particular, we focus on three different but related formalisms facilitating modeling of signaling processes with different levels of detail: interaction graphs, logical/Boolean networks, and logic-based ordinary differential equations (ODEs). Albeit the simplest models possible, interaction graphs allow the identification of important network properties such as signaling paths, feedback loops, or global interdependencies. Logical or Boolean models can be derived from interaction graphs by constraining the logical combination of edges. Logical models can be used to study the basic input-output behavior of the system under investigation and to analyze its qualitative dynamic properties by discrete simulations. They also provide a suitable framework to identify proper intervention strategies enforcing or repressing certain behaviors. Finally, as a third formalism, Boolean networks can be transformed into logic-based ODEs enabling studies on essential quantitative and dynamic features of a signaling network, where time and states are continuous.We describe and illustrate key methods and applications of the different modeling formalisms and discuss their relationships. In particular, as one important aspect for model reuse, we will show how these three modeling approaches can be combined to a modeling pipeline (or model hierarchy) allowing one to start with the simplest representation of a signaling network (interaction graph), which can later be refined to logical and eventually to logic-based ODE models. Importantly, systems and network properties determined in the rougher representation are conserved during these transformations.

Modeling innate immune response to early Mycobacterium infection

In the study of complex patterns in biology, mathematical and computational models are emerging as important tools. In addition to experimental approaches, these modeling tools have recently been applied to address open questions regarding host-pathogen interaction dynamics, including the immune response to mycobacterial infection and tuberculous granuloma formation. We present an approach in which a computational model represents the interaction of the Mycobacterium infection with the innate immune system in zebrafish at a high level of abstraction. We use the Petri Net formalism to model the interaction between the key host elements involved in granuloma formation and infection dissemination. We define a qualitative model for the understanding and description of causal relations in this dynamic process. Complex processes involving cell-cell or cell-bacteria communication can be modeled at smaller scales and incorporated hierarchically into this main model; these are to be included in later elaborations. With the infection mechanism being defined on a higher level, lower-level processes influencing the host-pathogen interaction can be identified, modeled, and tested both quantitatively and qualitatively. This systems biology framework incorporates modeling to generate and test hypotheses, to perform virtual experiments, and to make experimentally verifiable predictions. Thereby it supports the unraveling of the mechanisms of tuberculosis infection.

From qualitative data to quantitative models: analysis of the phage shock protein stress response in Escherichia coli

Background

Bacteria have evolved a rich set of mechanisms for sensing and adapting to adverse conditions in their environment. These are crucial for their survival, which requires them to react to extracellular stresses such as heat shock, ethanol treatment or phage infection. Here we focus on studying the phage shock protein (Psp) stress response in Escherichia coli induced by a phage infection or other damage to the bacterial membrane. This system has not yet been theoretically modelled or analysed in silico.

Results

We develop a model of the Psp response system, and illustrate how such models can be constructed and analyzed in light of available sparse and qualitative information in order to generate novel biological hypotheses about their dynamical behaviour. We analyze this model using tools from Petri-net theory and study its dynamical range that is consistent with currently available knowledge by conditioning model parameters on the available data in an approximate Bayesian computation (ABC) framework. Within this ABC approach we analyze stochastic and deterministic dynamics. This analysis allows us to identify different types of behaviour and these mechanistic insights can in turn be used to design new, more detailed and time-resolved experiments.

Conclusions

We have developed the first mechanistic model of the Psp response in E. coli. This model allows us to predict the possible qualitative stochastic and deterministic dynamic behaviours of key molecular players in the stress response. Our inferential approach can be applied to stress response and signalling systems more generally: in the ABC framework we can condition mathematical models on qualitative data in order to delimit e.g. parameter ranges or the qualitative system dynamics in light of available end-point or qualitative information.

Some aspects of the anemia of chronic disorders modeled and analyzed by petri net based approach

Anemia of chronic disorders is a very important phenomenon and iron is a crucial factor of this complex process. To better understand this process and its influence on some other factors we have built a mathematical model of the human body iron homeostasis, which possibly most exactly would reflect the metabolism of iron in the case of anemia and inflammation. The model has been formulated in the language of Petri net theory, which allows for its simulation and precise analysis. The obtained results of the analysis of the model’s behavior, concerning the influence of anemia and inflammation on the transferrin receptors, and hepcidin concentration changes are the valuable complements to the knowledge following from clinical research. This analysis is one of the first attempts to investigate properties and behavior of a not fully understood biological system on a basis of its Petri net based model.

A mathematical framework for agent based models of complex biological networks

Agent-based modeling and simulation is a useful method to study biological phenomena in a wide range of fields, from molecular biology to ecology. Since there is currently no agreed-upon standard way to specify such models, it is not always easy to use published models. Also, since model descriptions are not usually given in mathematical terms, it is difficult to bring mathematical analysis tools to bear, so that models are typically studied through simulation. In order to address this issue, Grimm et al. proposed a protocol for model specification, the so-called ODD protocol, which provides a standard way to describe models. This paper proposes an addition to the ODD protocol which allows the description of an agent-based model as a dynamical system, which provides access to computational and theoretical tools for its analysis. The mathematical framework is that of algebraic models, that is, time-discrete dynamical systems with algebraic structure. It is shown by way of several examples how this mathematical specification can help with model analysis. This mathematical framework can also accommodate other model types such as Boolean networks and the more general logical models, as well as Petri nets.

Automatically extracting functionally equivalent proteins from SwissProt

Background

There is a frequent need to obtain sets of functionally equivalent homologous proteins (FEPs) from different species. While it is usually the case that orthology implies functional equivalence, this is not always true; therefore datasets of orthologous proteins are not appropriate. The information relevant to extracting FEPs is contained in databanks such as UniProtKB/Swiss-Prot and a manual analysis of these data allow FEPs to be extracted on a one-off basis. However there has been no resource allowing the easy, automatic extraction of groups of FEPs – for example, all instances of protein C.

We have developed FOSTA, an automatically generated database of FEPs annotated as having the same function in UniProtKB/Swiss-Prot which can be used for large-scale analysis. The method builds a candidate list of homologues and filters out functionally diverged proteins on the basis of functional annotations using a simple text mining approach.

Results

Large scale evaluation of our FEP extraction method is difficult as there is no gold-standard dataset against which the method can be benchmarked. However, a manual analysis of five protein families confirmed a high level of performance. A more extensive comparison with two manually verified functional equivalence datasets also demonstrated very good performance.

Conclusion

In summary, FOSTA provides an automated analysis of annotations in UniProtKB/Swiss-Prot to enable groups of proteins already annotated as functionally equivalent, to be extracted. Our results demonstrate that the vast majority of UniProtKB/Swiss-Prot functional annotations are of high quality, and that FOSTA can interpret annotations successfully. Where FOSTA is not successful, we are able to highlight inconsistencies in UniProtKB/Swiss-Prot annotation. Most of these would have presented equal difficulties for manual interpretation of annotations. We discuss limitations and possible future extensions to FOSTA, and recommend changes to the UniProtKB/Swiss-Prot format, which would facilitate text-mining of UniProtKB/Swiss-Prot.

Toward the automated generation of genome-scale metabolic networks in the SEED

Background

Current methods for the automated generation of genome-scale metabolic networks focus on genome annotation and preliminary biochemical reaction network assembly, but do not adequately address the process of identifying and filling gaps in the reaction network, and verifying that the network is suitable for systems level analysis. Thus, current methods are only sufficient for generating draft-quality networks, and refinement of the reaction network is still largely a manual, labor-intensive process.

Results

We have developed a method for generating genome-scale metabolic networks that produces substantially complete reaction networks, suitable for systems level analysis. Our method partitions the reaction space of central and intermediary metabolism into discrete, interconnected components that can be assembled and verified in isolation from each other, and then integrated and verified at the level of their interconnectivity. We have developed a database of components that are common across organisms, and have created tools for automatically assembling appropriate components for a particular organism based on the metabolic pathways encoded in the organism's genome. This focuses manual efforts on that portion of an organism's metabolism that is not yet represented in the database. We have demonstrated the efficacy of our method by reverse-engineering and automatically regenerating the reaction network from a published genome-scale metabolic model for Staphylococcus aureus. Additionally, we have verified that our method capitalizes on the database of common reaction network components created for S. aureus, by using these components to generate substantially complete reconstructions of the reaction networks from three other published metabolic models (Escherichia coli, Helicobacter pylori, and Lactococcus lactis). We have implemented our tools and database within the SEED, an open-source software environment for comparative genome annotation and analysis.

Conclusion

Our method sets the stage for the automated generation of substantially complete metabolic networks for over 400 complete genome sequences currently in the SEED. With each genome that is processed using our tools, the database of common components grows to cover more of the diversity of metabolic pathways. This increases the likelihood that components of reaction networks for subsequently processed genomes can be retrieved from the database, rather than assembled and verified manually.

Application of Petri net based analysis techniques to signal transduction pathways

BACKGROUND

Signal transduction pathways are usually modelled using classical quantitative methods, which are based on ordinary differential equations (ODEs). However, some difficulties are inherent in this approach. On the one hand, the kinetic parameters involved are often unknown and have to be estimated. With increasing size and complexity of signal transduction pathways, the estimation of missing kinetic data is not possible. On the other hand, ODEs based models do not support any explicit insights into possible (signal-) flows within the network. Moreover, a huge amount of qualitative data is available due to high-throughput techniques. In order to get information on the systems behaviour, qualitative analysis techniques have been developed. Applications of the known qualitative analysis methods concern mainly metabolic networks. Petri net theory provides a variety of established analysis techniques, which are also applicable to signal transduction models. In this context special properties have to be considered and new dedicated techniques have to be designed.

METHODS

We apply Petri net theory to model and analyse signal transduction pathways first qualitatively before continuing with quantitative analyses. This paper demonstrates how to build systematically a discrete model, which reflects provably the qualitative biological behaviour without any knowledge of kinetic parameters. The mating pheromone response pathway in Saccharomyces cerevisiae serves as case study.

RESULTS

We propose an approach for model validation of signal transduction pathways based on the network structure only. For this purpose, we introduce the new notion of feasible t-invariants, which represent minimal self-contained subnets being active under a given input situation. Each of these subnets stands for a signal flow in the system. We define maximal common transition sets (MCT-sets), which can be used for t-invariant examination and net decomposition into smallest biologically meaningful functional units.

CONCLUSION

The paper demonstrates how Petri net analysis techniques can promote a deeper understanding of signal transduction pathways. The new concepts of feasible t-invariants and MCT-sets have been proven to be useful for model validation and the interpretation of the biological system behaviour. Whereas MCT-sets provide a decomposition of the net into disjunctive subnets, feasible t-invariants describe subnets, which generally overlap. This work contributes to qualitative modelling and to the analysis of large biological networks by their fully automatic decomposition into biologically meaningful modules.

Metabolite coupling in genome-scale metabolic networks

Background

Biochemically detailed stoichiometric matrices have now been reconstructed for various bacteria, yeast, and for the human cardiac mitochondrion based on genomic and proteomic data. These networks have been manually curated based on legacy data and elementally and charge balanced. Comparative analysis of these well curated networks is now possible. Pairs of metabolites often appear together in several network reactions, linking them topologically. This co-occurrence of pairs of metabolites in metabolic reactions is termed herein "metabolite coupling." These metabolite pairs can be directly computed from the stoichiometric matrix, S. Metabolite coupling is derived from the matrix ŜŜ^T, whose off-diagonal elements indicate the number of reactions in which any two metabolites participate together, where Ŝ is the binary form of S.

Results

Metabolite coupling in the studied networks was found to be dominated by a relatively small group of highly interacting pairs of metabolites. As would be expected, metabolites with high individual metabolite connectivity also tended to be those with the highest metabolite coupling, as the most connected metabolites couple more often. For metabolite pairs that are not highly coupled, we show that the number of reactions a pair of metabolites shares across a metabolic network closely approximates a line on a log-log scale. We also show that the preferential coupling of two metabolites with each other is spread across the spectrum of metabolites and is not unique to the most connected metabolites. We provide a measure for determining which metabolite pairs couple more often than would be expected based on their individual connectivity in the network and show that these metabolites often derive their principal biological functions from existing in pairs. Thus, analysis of metabolite coupling provides information beyond that which is found from studying the individual connectivity of individual metabolites.

Conclusion

The coupling of metabolites is an important topological property of metabolic networks. By computing coupling quantitatively for the first time in genome-scale metabolic networks, we provide insight into the basic structure of these networks.

In silico discovery of enzyme-substrate specificity-determining residue clusters

The binding between an enzyme and its substrate is highly specific, despite the fact that many different enzymes show significant sequence and structure similarity. There must be, then, substrate specificity-determining residues that enable different enzymes to recognize their unique substrates. We reason that a coordinated, not independent, action of both conserved and non-conserved residues determine enzymatic activity and specificity. Here, we present a surface patch ranking (SPR) method for in silico discovery of substrate specificity-determining residue clusters by exploring both sequence conservation and correlated mutations. As case studies we apply SPR to several highly homologous enzymatic protein pairs, such as guanylyl versus adenylyl cyclases, lactate versus malate dehydrogenases, and trypsin versus chymotrypsin. Without using experimental data, we predict several single and multi-residue clusters that are consistent with previous mutagenesis experimental results. Most single-residue clusters are directly involved in enzyme-substrate interactions, whereas multi-residue clusters are vital for domain-domain and regulator-enzyme interactions, indicating their complementary role in specificity determination. These results demonstrate that SPR may help the selection of target residues for mutagenesis experiments and, thus, focus rational drug design, protein engineering, and functional annotation to the relevant regions of a protein.

Knowledge-based voting algorithm for automated protein functional annotation†

Automated annotation of high-throughput genome sequences is one of the earliest steps toward a comprehensive understanding of the dynamic behavior of living organisms. However, the step is often error-prone because of its underlying algorithms, which rely mainly on a simple similarity analysis, and lack of guidance from biological rules. We present herein a knowledge-based protein annotation algorithm. Our objectives are to reduce errors and to improve annotation confidences. This algorithm consists of two major components: a knowledge system, called "RuleMiner," and a voting procedure. The knowledge system, which includes biological rules and functional profiles for each function, provides a platform for seamless integration of multiple sequence analysis tools and guidance for function annotation. The voting procedure, which relies on the knowledge system, is designed to make (possibly) unbiased judgments in functional assignments among complicated, sometimes conflicting, information. We have applied this algorithm to 10 prokaryotic bacterial genomes and observed a significant improvement in annotation confidences. We also discuss the current limitations of the algorithm and the potential for future improvement.