Biology subjects

Banga, J. R.

Publications and source records attributed to Banga, J. R..

11 recordsLinked to original sources

bioRxiv2026

Large-scale analysis of optimisation methods for parameter estimation problems in the life sciences

Dynamic models are central to the computational life sciences but typically contain unknown parameters that must be inferred from experimental data. High-throughput measurements have made this task increasingly challenging, yielding high-dimensional search spaces and non-convex objectives with many local optima. This makes the choice of optimisation method critical. However, existing empirical studies either consider only a limited number of benchmark problems or only a narrow spectrum of local, global and hybrid optimisation methods. Here, we present a comprehensive benchmark of a broad range of optimisation methods on a curated collection of parameter estimation problems, comprising 990 method-problem-pairs executed on two independent supercomputing infrastructures. Our evaluation quantifies success rates, solution quality and computational cost, revealing characteristic strengths and limitations of each approach. We find that optimisation methods separated into clear performance tiers. Building on these results, we implemented a new hybrid strategy that combines enhanced scatter search with the best-performing local solver, which showed robust performance and improved on the other scatter-search variants we tested. Our results provide practical guidance for selecting optimisation methods and thereby support more accurate and reliable model calibration.

systems biology↗

bioRxiv2026

An adaptive multi-strategy metaheuristic for robust model calibration in large-scale systems biology

Parameter estimation is a central challenge in systems biology, particularly for large dynamic models described by nonlinear ordinary differential equations (ODEs). These global optimization problems exhibit landscapes which are topologically heterogeneous, often exhibiting a pathological mixture of stiff, smooth valleys and rugged, noisy plateaus, making single-strategy hybrids ineffective. While methods like enhanced Scatter Search (eSS) represent the current state-of-the-art, their rigid intensification strategies can limit performance in large-scale or ill-conditioned scenarios. In this work, we introduce eLSHADE+, a novel metaheuristic architecture designed to adapt to these topological challenges. The proposed algorithm augments a recent Differential Evolution variant (LSHADE) with a probabilistic multistrategy hybridization. Unlike standard memetic algorithms, eLSHADE+ uses three distinct operational modes: (i) gradient-based intensification for precision in differentiable regions, (ii) derivative-free search for robustness against numerical noise, and (iii) pure global exploration to conserve computational resources in complicated basins. Additionally, a logarithmic parameter space transformation is incorporated to facilitate search across multi-scale biological constants. We rigorously evaluated eLSHADE+ using the BioPreDyn benchmark suite, comprising challenging real-world problems. Comparative analysis demonstrates that this adaptive multi-strategy approach yields statistically superior convergence speed and solution accuracy compared to eSS and other competitive metaheuristics, establishing a new baseline for robust model calibration in computational biology. Code and data are available at https://doi.org/10.5281/zenodo.18379327.

systems biology↗

bioRxiv2026

Host-aware Identification of Intrinsic Gene Expression Biopart Parameters from Combinatorial Libraries

Model-based design in synthetic biology is limited by the lack of quantitative, mechanistically interpretable biopart parameters that remain valid across genetic and physiological contexts. This limitation is particularly acute for transcriptional units, whose expression phenotypes emerge from nonlinear coupling between plasmid copy number, transcription, translation, and host resource allocation. Here we introduce a context- and host-aware framework for the absolute characterisation of gene expression bioparts embedded in combinatorial libraries of constitutive transcriptional units. Our approach leverages a digital twin of Escherichia coli conditioned on experimentally measured growth rate, used as a low-dimensional physiological proxy for cellular state. By embedding this growth-conditioned digital twin in a model-in-the-loop identification strategy, host-circuit interactions are explicitly accounted for and decoupled from intrinsic biopart properties. Using structured combinatorial libraries, we identify biophysically interpretable and transferable parameters for plasmid origins, promoters, and ribosome binding sites. In particular, we uncover an intrinsic translation initiation capacity of ribosome binding sites that remains invariant across genetic contexts and growth conditions, while context-dependent translation rates emerge as physiological projections of this invariant descriptor. This intrinsic parameterisation enables accurate prediction of protein synthesis across diverse host states, supports incremental and patchwork library expansion, and reveals localized failures of modularity that are obscured by phenotype-only characterisation. Together, these results establish a principled link between DNA sequence, intrinsic biopart parameters, and circuit-level phenotypes, providing a scalable and host-aware foundation for predictive design in synthetic biology.

synthetic biology↗

bioRxiv2025

Identification of dynamic models of microbial communities

Microbial communities, complex ecological networks crucial for human and planetary health, remain poorly understood in terms of the quantitative principles governing their composition, assembly, and function. Dynamic modeling using ordinary differential equations (ODEs) is a powerful framework for understanding and predicting microbiome behaviors. However, developing reliable ODE models is severely hampered by their nonlinear nature and the presence of significant challenges, particularly critical issues related to identifiability. Here, we address the identification problem in dynamic microbial community models by proposing an integrated methodology to tackle key challenges. Focusing on nonlinear ODE-based models, we examine four critical pitfalls: identifiability issues (structural and practical), unstable dynamics (potentially leading to numerical blow-up), underfitting (convergence to suboptimal solutions), and overfitting (fitting noise rather than signal). These pitfalls yield unreliable parameter estimates, unrealistic model behavior, and poor generalization. Our study presents a comprehensive workflow incorporating structural and practical identifiability analysis, robust global optimization for calibration, stability checks, and rigorous predictive power assessment. The methodologys effectiveness and versatility in mitigating these pitfalls are demonstrated through case studies of increasing complexity, paving the way for more reliable and mechanistically insightful models of microbial communities. AvailabilityThe code that implements the methodology and reproduces the results is available at https://doi.org/10.5281/zenodo.15309438 Supplementary informationAdditional information supporting this manuscript is provided at https://doi.org/10.5281/zenodo.15309438. Author summaryMicrobial communities, vital for human and environmental health, are complex systems whose quantitative behaviors are not yet fully understood. Scientists employ mathematical models to study these communities, but developing accurate models of their dynamics is very challenging. Key difficulties include determining correct model parameters, ensuring model stability, and preventing the models from either under-learning from data or over-learning from noise. This paper presents a new, integrated methodology to overcome these obstacles. The approach provides a systematic workflow incorporating analyses of parameter identifiability, model stability, and predictive capabilities. By addressing these critical pitfalls, this research aims to facilitate the creation of more reliable and mechanistically insightful models, ultimately enhancing our understanding of microbial community dynamics.

systems biology↗

bioRxiv2023

Parameter estimation in a whole-brain network model of epilepsy: comparison of parallel global optimization solvers

The Virtual Epileptic Patient (VEP) refers to a computer-based representation of a patient with epilepsy that combines personalized anatomical data with dynamical models of abnormal brain activities. It is capable of generating spatio-temporal seizure patterns that resemble those recorded with invasive methods such as stereoelectro EEG data, allowing for the evaluation of clinical hypotheses before planning surgery. This study highlights the effectiveness of calibrating VEP models using a global optimization approach. The approach utilizes SaCeSS, a cooperative metaheuristic capable of parallel computation, to yield high-quality solutions without requiring excessive computational time. Through extensive benchmarking, our proposal successfully solved a set of different configurations of VEP models, demonstrating better scalability and superior performance against other parallel solvers. These results were further enhanced using a Bayesian optimization framework for hyperparameter tuning, with significant gains in terms of both accuracy and computational cost. Additionally, we added an scalable uncertainty quantification phase after model calibration, and used it to assess the variability in estimated parameters across different problems. Overall, this study has the potential to improve the estimation of pathological brain areas in drug-resistant epilepsy, thereby to inform the clinical decision-making process. Author summaryMotivated by the problem of parameter estimation in a set of whole-brain network models of epilepsy (of increasing complexity), this study addresses the question of choosing a robust global optimization solver that can be accelerated by exploiting parallelism in different infrastructures, from desktop workstations to supercomputers. By leveraging data-driven techniques with robust cooperative global optimization methods, we aim to achieve accurate parameter estimation with reduced reliance on prior information. This is due to the dependency of Bayesian inference on the level of information in the prior, while this approach allows us to quantify uncertainty in the absence of any prior knowledge effectively. In this work, we construct an efficient and accurate method to perform parameter estimation and uncertainty quantification for the VEP model, and we use it to infer the brain regional epileptogenicity from source and sensor level whole-brain data. Of specific interest is the ability of our method to produce inference for high-dimensional state-space models governed by deterministic, stochastic, well-behaved, and stiff differential equations, using only partial observations and sparse encoding from system states to the observation.

neuroscience↗

bioRxiv2023

Reusable rule-based cell cycle model explains compartment-resolved dynamics of 16 observables in RPE-1 cells

The mammalian cell cycle is regulated by a well-studied but complex biochemical reaction system. Computational models provide a particularly systematic and systemic description of the mechanisms governing mammalian cell cycle control. By combining both state-of-the-art multiplexed experimental methods and powerful computational tools, this work aims at improving on these models along four dimensions: model structure, validation data, validation methodology and model reusability. We developed a comprehensive model structure of the full cell cycle that qualitatively explains the behaviour of human retinal pigment epithelial-1 cells. To estimate the model parameters, time courses of eight cell cycle regulators in two compartments were reconstructed from single cell snapshot measurements. After optimisation with a parallel global optimisation metaheuristic we obtained excellent agreements between simulations and measurements. The PEtab specification of the optimisation problem facilitates reuse of model, data and/or optimisation results. Future perturbation experiments will improve parameter identifiability and allow for testing model predictive power. Such a predictive model may aid in drug discovery for cell cycle-related disorders. Author SummaryWhile there are numerous cell cycle models in the literature, mammalian cell cycle models typically suffer from four limitations. Firstly, the descriptions of biological mechanisms are often overly complicated yet insufficiently comprehensive and detailed. Secondly, there is a lack of experimental data to validate the model. Thirdly, inadequate parameter estimation procedures are used. Lastly, there is no standardized description of the model and/or optimization problem. To overcome these limitations, we combine best-in-class technology to address all four simultaneously. We use a rule-based model description to provide a concise and less error-prone representation of complex biology. By applying trajectory reconstruction algorithms to existing data from highly multiplexed immunofluorescence measurements, we obtained a rich dataset for model validation. Using a parallel global metaheuristic for parameter estimation allowed us to bring simulations and data in very good agreement. To maximize reproducibility and reusability of our work, the results are available in three popular formats: BioNetGen, SBML, and PEtab. Our model is generalizable to many healthy and transformed cell types. The PEtab specification of the optimization problem makes it straightforward to re-optimize the parameters for other cell lines. This may guide hypotheses on cell type-specific regulation of the cell cycle, potentially with clinical relevance.

systems biology↗

bioRxiv2023

Distilling identifiable and interpretable dynamic models from biological data

Mechanistic dynamical models allow us to study the behavior of complex biological systems. They can provide an objective and quantitative understanding that would be difficult to achieve through other means. However, the systematic development of these models is a non-trivial exercise and an open problem in computational biology. Currently, many research efforts are focused on model discovery, i.e. automating the development of interpretable models from data. One of the main frameworks is sparse regression, where the sparse identification of nonlinear dynamics (SINDy) algorithm and its variants have enjoyed great success. SINDy-PI is an extension which allows the discovery of rational nonlinear terms, thus enabling the identification of kinetic functions common in biochemical networks, such as Michaelis-Menten. SINDy-PI also pays special attention to the recovery of parsimonious models (Occams razor). Here we focus on biological models composed of sets of deterministic nonlinear ordinary differential equations. We present a methodology that, combined with SINDy-PI, allows the automatic discovery of structurally identifiable and observable models which are also mechanistically interpretable. The lack of structural identifiability and observability makes it impossible to uniquely infer parameter and state variables, which can compromise the usefulness of a model by distorting its mechanistic significance and hampering its ability to produce biological insights. We illustrate the performance of our method with six case studies. We find that, despite enforcing sparsity, SINDy-PI sometimes yields models that are unidentifiable. In these cases we show how our method transforms their equations in order to obtain a structurally identifiable and observable model which is also interpretable. Author summaryDynamical models provide a quantitative understanding of complex biological systems. Since their development is far from trivial, in recent years many research efforts focus on obtaining these models automatically from data. One of the most effective approaches is based on implicit sparse regression. This technique is able to infer biochemical networks with kinetic functions containing rational nonlinear terms. However, as we show here, one limitation is that it may yield models that are unidentifiable. These features may lead to inaccurate mechanistic interpretations and wrong biological insights. To overcome this limitation, we propose an integrated methodology that applies additional procedures in order to ensure that the discovered models are structurally identifiable, observable, and interpretable. We demonstrate our method with six challenging case studies of increasing model complexity.

systems biology↗

bioRxiv2023

Optimal evaluation of energy yield and driving force in microbial metabolic pathway variants

This work presents a methodology to evaluate the bioenergetic feasibility of alternative metabolic pathways for a given microbial conversion, optimising their energy yield and driving forces as a function of the concentration of metabolic intermediates. The tool, based on thermodynamic principles and multi-objective optimisation, accounts for pathway variants in terms of different electron carriers, as well as energy conservation (proton translocating) reactions within the pathway. The method also accommodates other constraints, some of them non-linear, such as the balance of conserved moieties. The approach involves the transformation of the maximum energy yield problem into a multi-objective linear optimisation problem which is then subsequently solved using the epsilon-constraint method, highlighting the trade-off between yield and rate in metabolic reactions. The methodology is applied to analyse several pathway alternatives occurring during propionate oxidation in anaerobic fermentation processes, as well as to the reverse TCA cycle pathway occurring during autotrophic microbial CO2 fixation. The results obtained using the developed methodology match previously reported literature, and bring about insights into the studied pathways.

systems biology↗

bioRxiv2022

Automated design of gene circuits with optimal mushroom-bifurcation behaviour

Recent advances in synthetic biology are enabling exciting technologies, including the next generation of biosensors, the rational design of cell memory, modulated synthetic cell differentiation and generic multi-functional bio-circuits. These novel applications require the design of gene circuits leading to sophisticated behaviours and functionalities. At the same time, designs need to be kept minimal to avoid compromising cell viability. Bifurcation theory of dynamical systems provides powerful tools to address complex nonlinear dynamics and multifunctionality, linking model topology and kinetic parameters with circuit behaviour. However, the challenge of incorporating bifurcation analysis to automated design has not been accomplished so far. In this work we present an optimisation-based method for the automated forward design of synthetic gene circuits with specified bifurcation diagrams, allowing us to find minimal topologies optimizing the required functionalities and taking into account additional requirements and/or context specifications. We apply the method to design of gene circuits exhibiting the so called mushroom bifurcation, a relatively unexplored multi-functional behaviour of particular relevance for developmental biology. Using the results of the optimisation analysis we explore the capabilities of the resulting circuits for possible applications in advanced biosensors, memory devices, and synthetic cell differentiation.

systems biology↗

bioRxiv2020

Using optimal control to understand complex metabolic pathways

BackgroundWe revisit the idea of explaining and predicting dynamics in biochemical pathways from first-principles. A promising approach is to exploit optimality principles that can be justified from an evolutionary perspective. In the context of the cell, several previous studies have explained the dynamics of simple metabolic pathways exploiting optimality principles in combination with dynamic models, i.e. using an optimal control framework. For example, dynamics of gene expression in small metabolic models can be explained assuming that cells have developed optimal adaptation strategies. Most of these works have considered rather simplified representations, such as small linear pathways, or reduced networks with a single branching point. ResultsHere we consider the extension of this approach to more realistic scenarios, i.e. biochemical pathways of arbitrary size and structure. We first show that exploiting optimality principles for these networks poses great challenges due to the complexity of the associated optimal control problems. Second, in order to surmount such challenges, we present a computational framework based on multicriteria optimal control which has been designed with scalability and efficiency in mind, extending several recent methods. This framework includes mechanisms to avoid common pitfalls, such as local optima, unstable solutions or excessive computation time. We illustrate its performance with several case studies considering the central carbon metabolism of S. cerevisiae and B. subtilis. In particular, we consider metabolic dynamics during nutrient shift experiments. ConclusionsWe show how multi-objective optimal control can be used to predict temporal profiles of enzyme activation and metabolite concentrations in complex metabolic pathways. Further, we show how the multicriteria approach allows us to consider general cost/benefit trade-offs that have been likely favored by evolution. In this study we have considered metabolic pathways, but this computational framework can also be applied to analyze the dynamics of other complex pathways, such as signal transduction networks.

systems biology↗

bioRxiv2018

Benchmarking optimization methods for parameter estimation in large kinetic models

MotivationMechanistic kinetic models usually contain unknown parameters, which need to be estimated by optimizing the fit of the model to experimental data. This task can be computationally challenging due to the presence of local optima and ill-conditioning. While a variety of optimization methods have been suggested to surmount these issues, it is not obvious how to choose the best one for a given problem a priori, since many factors can influence their performance. A systematic comparison of methods that are suited to parameter estimation problems of sizes ranging from tens to hundreds of optimization variables is currently missing, and smaller studies indeed provided contradictory findings.\n\nResultsHere, we use a collection of benchmark problems to evaluate the performance of two families of optimization methods: (i) a multi-start of deterministic local searches; and (ii) a hybrid metaheuristic combining stochastic global search with deterministic local searches. A fair comparison is ensured through a collaborative evaluation, involving researchers applying each method on a daily basis, and a consideration of multiple performance metrics capturing the trade-off between computational efficiency and robustness. Our results show that, thanks to recent advances in the calculation of parametric sensitivities, a multi-start of gradient-based local methods is often a successful strategy, but a better performance can be obtained with a hybrid metaheuristic. The best performer is a combination of a global scatter search metaheuristic with an interior point local method, provided with gradients estimated with adjoint-based sensitivities. We provide an implementation of this novel method in an open-source software toolbox to render it available to the scientific community.\n\nAvailability and ImplementationThe code to reproduce the results is available at Zenodo https://doi.org/10.5281/zenodo.1160343\n\nContactjan.hasenauer@helmholtz-muenchen.de, julio@iim.csic.es

systems biology↗