David Henriques

MEIGO: an open-source software suite based on metaheuristics for global optimization in systems biology and bioinformatics.

BackgroundOptimization is the key to solving many problems in computational biology. Global optimization methods, which provide a robust methodology, and metaheuristics in particular have proven to be the most efficient methods for many applications. Despite their utility, there is a limited availability of metaheuristic tools.ResultsWe present MEIGO, an R and Matlab optimization toolbox (also available in Python via a wrapper of the R version), that implements metaheuristics capable of solving diverse problems arising in systems biology and bioinformatics. The toolbox includes the enhanced scatter search method (eSS) for continuous nonlinear programming (cNLP) and mixed-integer programming (MINLP) problems, and variable neighborhood search (VNS) for Integer Programming (IP) problems. Additionally, the R version includes BayesFit for parameter estimation by Bayesian inference. The eSS and VNS methods can be run on a single-thread or in parallel using a cooperative strategy. The code is supplied under GPLv3 and is available at http://www.iim.csic.es/~gingproc/meigo.html. Documentation and examples are included. The R package has been submitted to BioConductor. We evaluate MEIGO against optimization benchmarks, and illustrate its applicability to a series of case studies in bioinformatics and systems biology where it outperforms other state-of-the-art methods.ConclusionsMEIGO provides a free, open-source platform for optimization that can be applied to multiple domains of systems biology and bioinformatics. It includes efficient state of the art metaheuristics, and its open and modular structure allows the addition of further methods.

State–time spectrum of signal transduction logic models

Despite the current wealth of high-throughput data, our understanding of signal transduction is still incomplete. Mathematical modeling can be a tool to gain an insight into such processes. Detailed biochemical modeling provides deep understanding, but does not scale well above relatively a few proteins. In contrast, logic modeling can be used where the biochemical knowledge of the system is sparse and, because it is parameter free (or, at most, uses relatively a few parameters), it scales well to large networks that can be derived by manual curation or retrieved from public databases. Here, we present an overview of logic modeling formalisms in the context of training logic models to data, and specifically the different approaches to modeling qualitative to quantitative data (state) and dynamics (time) of signal transduction. We use a toy model of signal transduction to illustrate how different logic formalisms (Boolean, fuzzy logic and differential equations) treat state and time. Different formalisms allow for different features of the data to be captured, at the cost of extra requirements in terms of computational power and data quality and quantity. Through this demonstration, the assumptions behind each formalism are discussed, as well as their advantages and disadvantages and possible future developments.

BioPreDyn-bench: a suite of benchmark problems for dynamic modelling in systems biology.

BackgroundDynamic modelling is one of the cornerstones of systems biology. Many research efforts are currently being invested in the development and exploitation of large-scale kinetic models. The associated problems of parameter estimation (model calibration) and optimal experimental design are particularly challenging. The community has already developed many methods and software packages which aim to facilitate these tasks. However, there is a lack of suitable benchmark problems which allow a fair and systematic evaluation and comparison of these contributions.ResultsHere we present BioPreDyn-bench, a set of challenging parameter estimation problems which aspire to serve as reference test cases in this area. This set comprises six problems including medium and large-scale kinetic models of the bacterium E. coli, baker’s yeast S. cerevisiae, the vinegar fly D. melanogaster, Chinese Hamster Ovary cells, and a generic signal transduction network. The level of description includes metabolism, transcription, signal transduction, and development. For each problem we provide (i) a basic description and formulation, (ii) implementations ready-to-run in several formats, (iii) computational results obtained with specific solvers, (iv) a basic analysis and interpretation.ConclusionsThis suite of benchmark problems can be readily used to evaluate and compare parameter estimation methods. Further, it can also be used to build test problems for sensitivity and identifiability analysis, model reduction and optimal experimental design methods. The suite, including codes and documentation, can be freely downloaded from the BioPreDyn-bench website, https://sites.google.com/site/biopredynbenchmarks/.

Reverse engineering of logic-based differential equation models using a mixed-integer dynamic optimization approach

Motivation: Systems biology models can be used to test new hypotheses formulated on the basis of previous knowledge or new experimental data, contradictory with a previously existing model. New hypotheses often come in the shape of a set of possible regulatory mechanisms. This search is usually not limited to finding a single regulation link, but rather a combination of links subject to great uncertainty or no information about the kinetic parameters. Results: In this work, we combine a logic-based formalism, to describe all the possible regulatory structures for a given dynamic model of a pathway, with mixed-integer dynamic optimization (MIDO). This framework aims to simultaneously identify the regulatory structure (represented by binary parameters) and the real-valued parameters that are consistent with the available experimental data, resulting in a logic-based differential equation model. The alternative to this would be to perform real-valued parameter estimation for each possible model structure, which is not tractable for models of the size presented in this work. The performance of the method presented here is illustrated with several case studies: a synthetic pathway problem of signaling regulation, a two-component signal transduction pathway in bacterial homeostasis, and a signaling network in liver cancer cells. Supplementary information: Supplementary data are available at Bioinformatics online. Contact: julio@iim.csic.es or saezrodriguez@ebi.ac.uk

Data-driven reverse engineering of signaling pathways using ensembles of dynamic models

Despite significant efforts and remarkable progress, the inference of signaling networks from experimental data remains very challenging. The problem is particularly difficult when the objective is to obtain a dynamic model capable of predicting the effect of novel perturbations not considered during model training. The problem is ill-posed due to the nonlinear nature of these systems, the fact that only a fraction of the involved proteins and their post-translational modifications can be measured, and limitations on the technologies used for growing cells in vitro, perturbing them, and measuring their variations. As a consequence, there is a pervasive lack of identifiability. To overcome these issues, we present a methodology called SELDOM (enSEmbLe of Dynamic lOgic-based Models), which builds an ensemble of logic-based dynamic models, trains them to experimental data, and combines their individual simulations into an ensemble prediction. It also includes a model reduction step to prune spurious interactions and mitigate overfitting. SELDOM is a data-driven method, in the sense that it does not require any prior knowledge of the system: the interaction networks that act as scaffolds for the dynamic models are inferred from data using mutual information. We have tested SELDOM on a number of experimental and in silico signal transduction case-studies, including the recent HPN-DREAM breast cancer challenge. We found that its performance is highly competitive compared to state-of-the-art methods for the purpose of recovering network topology. More importantly, the utility of SELDOM goes beyond basic network inference (i.e. uncovering static interaction networks): it builds dynamic (based on ordinary differential equation) models, which can be used for mechanistic interpretations and reliable dynamic predictions in new experimental conditions (i.e. not used in the training). For this task, SELDOM’s ensemble prediction is not only consistently better than predictions from individual models, but also often outperforms the state of the art represented by the methods used in the HPN-DREAM challenge.

Systems Pharmacology Dissection of Cholesterol Regulation Reveals Determinants of Large Pharmacodynamic Variability between Cell Lines

Summary In individuals, heterogeneous drug-response phenotypes result from a complex interplay of dose, drug specificity, genetic background, and environmental factors, thus challenging our understanding of the underlying processes and optimal use of drugs in the clinical setting. Here, we use mass-spectrometry-based quantification of molecular response phenotypes and logic modeling to explain drug-response differences in a panel of cell lines. We apply this approach to cellular cholesterol regulation, a biological process with high clinical relevance. From the quantified molecular phenotypes elicited by various targeted pharmacologic or genetic treatments, we generated cell-line-specific models that quantified the processes beneath the idiotypic intracellular drug responses. The models revealed that, in addition to drug uptake and metabolism, further cellular processes displayed significant pharmacodynamic response variability between the cell lines, resulting in cell-line-specific drug-response phenotypes. This study demonstrates the importance of integrating different types of quantitative systems-level molecular measurements with modeling to understand the effect of pharmacological perturbations on complex biological processes.

Modeling Signaling Networks with Different Formalisms: A Preview

In the last 30 years, many of the mechanisms behind signal transduction, the process by which the cell takes extracellular signals as an input and converts them to a specific cellular phenotype, have been experimentally determined. With these discoveries, however, has come the realization that the architecture of signal transduction, the signaling network, is incredibly complex. Although the main pathways between receptor and output are well-known, there is a complex net of regulatory features that include crosstalk between different pathways, spatial and temporal effects, and positive and negative feedbacks. Hence, modeling approaches have been used to try and unravel some of these complexities. We use the mitogen-activated protein kinase cascade to illustrate chemical kinetic and logic approaches to modeling signaling networks. By using a common well-known model, we illustrate here the assumptions and level of detail behind each modeling approach, which serves as an introduction to the more detailed discussions of each in the accompanying chapters in this book.

Saccharomyces cerevisiae and S. kudriavzevii Synthetic Wine Fermentation Performance Dissected by Predictive Modeling

Wineries face unprecedented challenges due to new market demands and climate change effects on wine quality. New yeast starters including non-conventional Saccharomyces species, such as S. kudriavzevii, may contribute to deal with some of these challenges. The design of new fermentations using non-conventional yeasts requires an improved understanding of the physiology and metabolism of these cells. Dynamic modeling brings the potential of exploring the most relevant mechanisms and designing optimal processes more systematically. In this work we explore mechanisms by means of a model selection, reduction and cross-validation pipeline which enables to dissect the most relevant fermentation features for the species under consideration, Saccharomyces cerevisiae T73 and Saccharomyces kudriavzevii CR85. The pipeline involved the comparison of a collection of models which incorporate several alternative mechanisms with emphasis on the inhibitory effects due to temperature and ethanol. We focused on defining a minimal model with the minimum number of parameters, to maximize the identifiability and the quality of cross-validation. The selected model was then used to highlight differences in behavior between species. The analysis of model parameters would indicate that the specific growth rate and the transport of hexoses at initial times are higher for S. cervisiae T73 while S. kudriavzevii CR85 diverts more flux for glycerol production and cellular maintenance. As a result, the fermentations with S. kudriavzevii CR85 are typically slower; produce less ethanol but higher glycerol. Finally, we also explored optimal initial inoculation and process temperature to find the best compromise between final product characteristics and fermentation duration. Results reveal that the production of glycerol is distinctive in S. kudriavzevii CR85, it was not possible to achieve the same production of glycerol with S. cervisiae T73 in any of the conditions tested. This result brings the idea that the optimal design of mixed cultures may have an enormous potential for the improvement of final wine quality.

A parallel metaheuristic for large mixedinteger dynamic optimization problems, with applications in computational biology

Background We consider a general class of global optimization problems dealing with nonlinear dynamic models. Although this class is relevant to many areas of science and engineering, here we are interested in applying this framework to the reverse engineering problem in computational systems biology, which yields very large mixed-integer dynamic optimization (MIDO) problems. In particular, we consider the framework of logic-based ordinary differential equations (ODEs). Methods We present saCeSS2, a parallel method for the solution of this class of problems. This method is based on an parallel cooperative scatter search metaheuristic, with new mechanisms of self-adaptation and specific extensions to handle large mixed-integer problems. We have paid special attention to the avoidance of convergence stagnation using adaptive cooperation strategies tailored to this class of problems. Results We illustrate its performance with a set of three very challenging case studies from the domain of dynamic modelling of cell signaling. The simpler case study considers a synthetic signaling pathway and has 84 continuous and 34 binary decision variables. A second case study considers the dynamic modeling of signaling in liver cancer using high-throughput data, and has 135 continuous and 109 binaries decision variables. The third case study is an extremely difficult problem related with breast cancer, involving 690 continuous and 138 binary decision variables. We report computational results obtained in different infrastructures, including a local cluster, a large supercomputer and a public cloud platform. Interestingly, the results show how the cooperation of individual parallel searches modifies the systemic properties of the sequential algorithm, achieving superlinear speedups compared to an individual search (e.g. speedups of 15 with 10 cores), and significantly improving (above a 60%) the performance with respect to a non-cooperative parallel scheme. The scalability of the method is also good (tests were performed using up to 300 cores). Conclusions These results demonstrate that saCeSS2 can be used to successfully reverse engineer large dynamic models of complex biological pathways. Further, these results open up new possibilities for other MIDO-based large-scale applications in the life sciences such as metabolic engineering, synthetic biology, drug scheduling.