Shenghao Yao, AmirHosein Sadeghimanesh, Matthew England
We present the first use of machine learning tools to predict multistationarity of reaction networks. Chemical Reaction Networks (CRNs) are the mathematical formulation of how the quantities associated to a set of species (molecules, proteins, cells, or animals) vary as time passes with respect to their interactions with each other. Their mathematics does not describe just chemical reactions but many other areas of the life sciences such as ecology, epidemiology, and population dynamics. We say a CRN is at a steady state when the concentration (or number) of species do not vary anymore. Some CRNs do not attain a steady state while some others may have more than one possible steady state. The CRNs in the later group are called multistationary. Multistationarity is an important property, e.g. switch-like behaviour in cells needs multistationarity to occur. Existing algorithms to detect whether a CRN is multistationary or not are either extremely expensive or restricted in the type of CRNs they can be used on, motivating a new machine learning approach. We address the problem of representing variable-length CRN data to machine learning models by developing a new graph representation of CRNs for use with graph learning algorithms. We contribute a large dataset of labelled fully open CRNs whose production necessitated the development of new CRN theory. Then we present experimental results on the training and testing of a graph attention network model on this dataset, showing excellent levels of performance. We finish by testing the model predictions on validation data produced independently, demonstrating generalisability of the model to different types of CRN.
{"title":"Designing Machine Learning Tools to Characterize Multistationarity of Fully Open Reaction Networks","authors":"Shenghao Yao, AmirHosein Sadeghimanesh, Matthew England","doi":"arxiv-2407.01760","DOIUrl":"https://doi.org/arxiv-2407.01760","url":null,"abstract":"We present the first use of machine learning tools to predict\u0000multistationarity of reaction networks. Chemical Reaction Networks (CRNs) are the mathematical formulation of how the\u0000quantities associated to a set of species (molecules, proteins, cells, or\u0000animals) vary as time passes with respect to their interactions with each\u0000other. Their mathematics does not describe just chemical reactions but many\u0000other areas of the life sciences such as ecology, epidemiology, and population\u0000dynamics. We say a CRN is at a steady state when the concentration (or number)\u0000of species do not vary anymore. Some CRNs do not attain a steady state while\u0000some others may have more than one possible steady state. The CRNs in the later\u0000group are called multistationary. Multistationarity is an important property,\u0000e.g. switch-like behaviour in cells needs multistationarity to occur. Existing\u0000algorithms to detect whether a CRN is multistationary or not are either\u0000extremely expensive or restricted in the type of CRNs they can be used on,\u0000motivating a new machine learning approach. We address the problem of representing variable-length CRN data to machine\u0000learning models by developing a new graph representation of CRNs for use with\u0000graph learning algorithms. We contribute a large dataset of labelled fully open\u0000CRNs whose production necessitated the development of new CRN theory. Then we\u0000present experimental results on the training and testing of a graph attention\u0000network model on this dataset, showing excellent levels of performance. We\u0000finish by testing the model predictions on validation data produced\u0000independently, demonstrating generalisability of the model to different types\u0000of CRN.","PeriodicalId":501325,"journal":{"name":"arXiv - QuanBio - Molecular Networks","volume":"21 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141518267","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
José E. R. Cury, Patrícia Tenera Roxo, Vasco Manquinho, Claudine Chaouiya, Pedro T. Monteiro
Boolean networks constitute relevant mathematical models to study the behaviours of genetic and signalling networks. These networks define regulatory influences between molecular nodes, each being associated to a Boolean variable and a regulatory (local) function specifying its dynamical behaviour depending on its regulators. However, existing data is mostly insufficient to adequately parametrise a model, that is to uniquely define a regulatory function for each node. With the intend to support model parametrisation, this paper presents results on the set of Boolean functions compatible with a given regulatory structure, i.e. the partially ordered set of monotone non-degenerate Boolean functions. More precisely, we present original rules to obtain the direct neighbours of any function of this set. Besides a theoretical interest, presented results will enable the development of more efficient methods for Boolean network synthesis and revision, benefiting from the progressive exploration of the vicinity of regulatory functions.
{"title":"Immediate Neighbours of Monotone Boolean Functions","authors":"José E. R. Cury, Patrícia Tenera Roxo, Vasco Manquinho, Claudine Chaouiya, Pedro T. Monteiro","doi":"arxiv-2407.01337","DOIUrl":"https://doi.org/arxiv-2407.01337","url":null,"abstract":"Boolean networks constitute relevant mathematical models to study the\u0000behaviours of genetic and signalling networks. These networks define regulatory\u0000influences between molecular nodes, each being associated to a Boolean variable\u0000and a regulatory (local) function specifying its dynamical behaviour depending\u0000on its regulators. However, existing data is mostly insufficient to adequately\u0000parametrise a model, that is to uniquely define a regulatory function for each\u0000node. With the intend to support model parametrisation, this paper presents\u0000results on the set of Boolean functions compatible with a given regulatory\u0000structure, i.e. the partially ordered set of monotone non-degenerate Boolean\u0000functions. More precisely, we present original rules to obtain the direct\u0000neighbours of any function of this set. Besides a theoretical interest,\u0000presented results will enable the development of more efficient methods for\u0000Boolean network synthesis and revision, benefiting from the progressive\u0000exploration of the vicinity of regulatory functions.","PeriodicalId":501325,"journal":{"name":"arXiv - QuanBio - Molecular Networks","volume":"22 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141503778","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Hugo BuscemiENS Paris Saclay, Lifeware, François FagesLifeware
Complex systems can be advantageously modeled by formal reaction systems (RS), a.k.a. chemical reaction networks in chemistry. Reaction-based models can indeed be interpreted in a hierarchy of semantics, depending on the question at hand, most notably by Ordinary Differential Equations (ODEs), Continuous Time Markov Chains (CTMCs), discrete Petri nets and asynchronous Boolean transition systems. The last three semantics can be easily related in the framework of abstract interpretation. The first two are classically related by Kurtz's limit theorem which states that if reactions are density-dependent families, then, as the volume goes to infinity, the mean reactant concentrations of the CTMC tends towards the solution of the ODE. In the more realistic context of bounded volumes, it is easy to show, by moment closure, that the restriction to reactions with at most one reactant ensures similarly that the mean of the CTMC trajectories is equal to the solution of the ODE at all time points. In this paper, we generalize that result in presence of polyreactant reactions, by introducing the Stoichiometric Influence and Modification Graph (SIMG) of an RS, and by showing that the equality between the two interpretations holds for the variables that belong to distinct SIMG ancestors of polyreactant reactions. We illustrate this approach with several examples. Evaluation on BioModels reveals that the condition for all variables is satisfied on models with no polymolecular reaction only. However, our theorem can be applied selectively to certain variables of the model to provide insights into their behaviour within more complex systems. Interestingly, we also show that the equality holds for a basic oscillatory RS implementing the sine and cosine functions of time.
{"title":"Graphical Conditions ensuring Equality between Differential and Mean Stochastic Dynamics","authors":"Hugo BuscemiENS Paris Saclay, Lifeware, François FagesLifeware","doi":"arxiv-2406.18126","DOIUrl":"https://doi.org/arxiv-2406.18126","url":null,"abstract":"Complex systems can be advantageously modeled by formal reaction systems\u0000(RS), a.k.a. chemical reaction networks in chemistry. Reaction-based models can\u0000indeed be interpreted in a hierarchy of semantics, depending on the question at\u0000hand, most notably by Ordinary Differential Equations (ODEs), Continuous Time\u0000Markov Chains (CTMCs), discrete Petri nets and asynchronous Boolean transition\u0000systems. The last three semantics can be easily related in the framework of\u0000abstract interpretation. The first two are classically related by Kurtz's limit\u0000theorem which states that if reactions are density-dependent families, then, as\u0000the volume goes to infinity, the mean reactant concentrations of the CTMC tends\u0000towards the solution of the ODE. In the more realistic context of bounded\u0000volumes, it is easy to show, by moment closure, that the restriction to\u0000reactions with at most one reactant ensures similarly that the mean of the CTMC\u0000trajectories is equal to the solution of the ODE at all time points. In this\u0000paper, we generalize that result in presence of polyreactant reactions, by\u0000introducing the Stoichiometric Influence and Modification Graph (SIMG) of an\u0000RS, and by showing that the equality between the two interpretations holds for\u0000the variables that belong to distinct SIMG ancestors of polyreactant reactions.\u0000We illustrate this approach with several examples. Evaluation on BioModels\u0000reveals that the condition for all variables is satisfied on models with no\u0000polymolecular reaction only. However, our theorem can be applied selectively to\u0000certain variables of the model to provide insights into their behaviour within\u0000more complex systems. Interestingly, we also show that the equality holds for a\u0000basic oscillatory RS implementing the sine and cosine functions of time.","PeriodicalId":501325,"journal":{"name":"arXiv - QuanBio - Molecular Networks","volume":"49 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141503779","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Elisa Gómez de Lope, Saurabh Deshpande, Ramón Viñas Torné, Pietro Liò, Enrico Glaab, Stéphane P. A. Bordas
Omics data analysis is crucial for studying complex diseases, but its high dimensionality and heterogeneity challenge classical statistical and machine learning methods. Graph neural networks have emerged as promising alternatives, yet the optimal strategies for their design and optimization in real-world biomedical challenges remain unclear. This study evaluates various graph representation learning models for case-control classification using high-throughput biological data from Parkinson's disease and control samples. We compare topologies derived from sample similarity networks and molecular interaction networks, including protein-protein and metabolite-metabolite interactions (PPI, MMI). Graph Convolutional Network (GCNs), Chebyshev spectral graph convolution (ChebyNet), and Graph Attention Network (GAT), are evaluated alongside advanced architectures like graph transformers, the graph U-net, and simpler models like multilayer perceptron (MLP). These models are systematically applied to transcriptomics and metabolomics data independently. Our comparative analysis highlights the benefits and limitations of various architectures in extracting patterns from omics data, paving the way for more accurate and interpretable models in biomedical research.
{"title":"Graph Representation Learning Strategies for Omics Data: A Case Study on Parkinson's Disease","authors":"Elisa Gómez de Lope, Saurabh Deshpande, Ramón Viñas Torné, Pietro Liò, Enrico Glaab, Stéphane P. A. Bordas","doi":"arxiv-2406.14442","DOIUrl":"https://doi.org/arxiv-2406.14442","url":null,"abstract":"Omics data analysis is crucial for studying complex diseases, but its high\u0000dimensionality and heterogeneity challenge classical statistical and machine\u0000learning methods. Graph neural networks have emerged as promising alternatives,\u0000yet the optimal strategies for their design and optimization in real-world\u0000biomedical challenges remain unclear. This study evaluates various graph\u0000representation learning models for case-control classification using\u0000high-throughput biological data from Parkinson's disease and control samples.\u0000We compare topologies derived from sample similarity networks and molecular\u0000interaction networks, including protein-protein and metabolite-metabolite\u0000interactions (PPI, MMI). Graph Convolutional Network (GCNs), Chebyshev spectral\u0000graph convolution (ChebyNet), and Graph Attention Network (GAT), are evaluated\u0000alongside advanced architectures like graph transformers, the graph U-net, and\u0000simpler models like multilayer perceptron (MLP). These models are systematically applied to transcriptomics and metabolomics\u0000data independently. Our comparative analysis highlights the benefits and\u0000limitations of various architectures in extracting patterns from omics data,\u0000paving the way for more accurate and interpretable models in biomedical\u0000research.","PeriodicalId":501325,"journal":{"name":"arXiv - QuanBio - Molecular Networks","volume":"24 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141503780","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yue Wang, Peng Zheng, Yu-Chen Cheng, Zikun Wang, Aleksandr Aravkin
Determining gene regulatory network (GRN) structure is a central problem in biology, with a variety of inference methods available for different types of data. For a widely prevalent and challenging use case, namely single-cell gene expression data measured after intervention at multiple time points with unknown joint distributions, there is only one known specifically developed method, which does not fully utilize the rich information contained in this data type. We develop an inference method for the GRN in this case, netWork infErence by covariaNce DYnamics, dubbed WENDY. The core idea of WENDY is to model the dynamics of the covariance matrix, and solve this dynamics as an optimization problem to determine the regulatory relationships. To evaluate its effectiveness, we compare WENDY with other inference methods using synthetic data and experimental data. Our results demonstrate that WENDY performs well across different data sets.
{"title":"Gene Regulatory Network Inference with Covariance Dynamics","authors":"Yue Wang, Peng Zheng, Yu-Chen Cheng, Zikun Wang, Aleksandr Aravkin","doi":"arxiv-2407.00754","DOIUrl":"https://doi.org/arxiv-2407.00754","url":null,"abstract":"Determining gene regulatory network (GRN) structure is a central problem in\u0000biology, with a variety of inference methods available for different types of\u0000data. For a widely prevalent and challenging use case, namely single-cell gene\u0000expression data measured after intervention at multiple time points with\u0000unknown joint distributions, there is only one known specifically developed\u0000method, which does not fully utilize the rich information contained in this\u0000data type. We develop an inference method for the GRN in this case, netWork\u0000infErence by covariaNce DYnamics, dubbed WENDY. The core idea of WENDY is to\u0000model the dynamics of the covariance matrix, and solve this dynamics as an\u0000optimization problem to determine the regulatory relationships. To evaluate its\u0000effectiveness, we compare WENDY with other inference methods using synthetic\u0000data and experimental data. Our results demonstrate that WENDY performs well\u0000across different data sets.","PeriodicalId":501325,"journal":{"name":"arXiv - QuanBio - Molecular Networks","volume":"52 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141518268","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The Hilbert number $H(n)$ is defined as the maximum number of limit cycles of a planar autonomous system of ordinary differential equations (ODEs) with right-hand sides containing polynomials of degree at most $n in {mathbb N}$. The dynamics of chemical reaction systems with two chemical species can be (under mass-action kinetics) described by such planar autonomous ODEs, where $n$ is equal to the maximum order of the chemical reactions in the system. Generalizations of the Hilbert number $H(n)$ to three different classes of chemical reaction networks are investigated: (i) chemical systems with reactions up to the $n$-th order; (ii) systems with up to $n$-molecular chemical reactions; and (iii) weakly reversible chemical reaction networks. In each case (i), (ii) and (iii), the question on the number of limit cycles is considered. Lower bounds on the generalized Hilbert numbers are provided for both algebraic and non-algebraic limit cycles. Furthermore, given a general algebraic curve $h(x,y)=0$ of degree $n_h in {mathbb N}$ and containing one or more ovals in the positive quadrant, a chemical system is constructed which has the oval(s) as its stable algebraic limit cycle(s). The ODEs describing the dynamics of the constructed chemical system contain polynomials of degree at most $n=2,n_h+1.$ Considering $n_h ge 4,$ the algebraic curve $h(x,y)=0$ can contain multiple closed components with the maximum number of ovals given by Harnack's curve theorem as $1+(n_h-1)(n_h-2)/2$, which is equal to 4 for $n_h=4.$ Algebraic curve $h(x,y)=0$ with $n_h=4$ and the maximum number of four ovals is used to construct a chemical system which has four stable algebraic limit cycles.
{"title":"Planar chemical reaction systems with algebraic and non-algebraic limit cycles","authors":"Gheorghe Craciun, Radek Erban","doi":"arxiv-2406.05057","DOIUrl":"https://doi.org/arxiv-2406.05057","url":null,"abstract":"The Hilbert number $H(n)$ is defined as the maximum number of limit cycles of\u0000a planar autonomous system of ordinary differential equations (ODEs) with\u0000right-hand sides containing polynomials of degree at most $n in {mathbb N}$.\u0000The dynamics of chemical reaction systems with two chemical species can be\u0000(under mass-action kinetics) described by such planar autonomous ODEs, where\u0000$n$ is equal to the maximum order of the chemical reactions in the system.\u0000Generalizations of the Hilbert number $H(n)$ to three different classes of\u0000chemical reaction networks are investigated: (i) chemical systems with\u0000reactions up to the $n$-th order; (ii) systems with up to $n$-molecular\u0000chemical reactions; and (iii) weakly reversible chemical reaction networks. In\u0000each case (i), (ii) and (iii), the question on the number of limit cycles is\u0000considered. Lower bounds on the generalized Hilbert numbers are provided for\u0000both algebraic and non-algebraic limit cycles. Furthermore, given a general\u0000algebraic curve $h(x,y)=0$ of degree $n_h in {mathbb N}$ and containing one\u0000or more ovals in the positive quadrant, a chemical system is constructed which\u0000has the oval(s) as its stable algebraic limit cycle(s). The ODEs describing the\u0000dynamics of the constructed chemical system contain polynomials of degree at\u0000most $n=2,n_h+1.$ Considering $n_h ge 4,$ the algebraic curve $h(x,y)=0$ can\u0000contain multiple closed components with the maximum number of ovals given by\u0000Harnack's curve theorem as $1+(n_h-1)(n_h-2)/2$, which is equal to 4 for\u0000$n_h=4.$ Algebraic curve $h(x,y)=0$ with $n_h=4$ and the maximum number of four\u0000ovals is used to construct a chemical system which has four stable algebraic\u0000limit cycles.","PeriodicalId":501325,"journal":{"name":"arXiv - QuanBio - Molecular Networks","volume":"25 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141503781","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Dynamical systems with polynomials on the right-hand side can model a wide range of physical processes. A subset of such dynamical systems that can model chemical reactions under mass-action kinetics are called chemical systems. A central problem in synthetic biology is to map general polynomial dynamical systems into dynamically similar chemical ones. In this paper, we present a novel map, called the quasi-chemical map, that can systematically solve this problem. The quasi-chemical map introduces suitable state-dependent perturbations into any given polynomial dynamical system which then becomes chemical under suitably large translation of variables. We prove that this map preserves robust dynamical features, such as generic equilibria and limit cycles, as well as temporal properties, such as periods of oscillations. Furthermore, the resulting chemical systems are of only at most one degree higher than the original dynamical systems. We demonstrate the quasi-chemical map by designing relatively simple chemical systems with exotic dynamics and pre-defined bifurcation structures.
{"title":"Mapping dynamical systems into chemical reactions","authors":"Tomislav Plesa","doi":"arxiv-2406.03473","DOIUrl":"https://doi.org/arxiv-2406.03473","url":null,"abstract":"Dynamical systems with polynomials on the right-hand side can model a wide\u0000range of physical processes. A subset of such dynamical systems that can model\u0000chemical reactions under mass-action kinetics are called chemical systems. A\u0000central problem in synthetic biology is to map general polynomial dynamical\u0000systems into dynamically similar chemical ones. In this paper, we present a\u0000novel map, called the quasi-chemical map, that can systematically solve this\u0000problem. The quasi-chemical map introduces suitable state-dependent\u0000perturbations into any given polynomial dynamical system which then becomes\u0000chemical under suitably large translation of variables. We prove that this map\u0000preserves robust dynamical features, such as generic equilibria and limit\u0000cycles, as well as temporal properties, such as periods of oscillations.\u0000Furthermore, the resulting chemical systems are of only at most one degree\u0000higher than the original dynamical systems. We demonstrate the quasi-chemical\u0000map by designing relatively simple chemical systems with exotic dynamics and\u0000pre-defined bifurcation structures.","PeriodicalId":501325,"journal":{"name":"arXiv - QuanBio - Molecular Networks","volume":"46 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141550854","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Alexander Dack, Benjamin Qureshi, Thomas E. Ouldridge, Tomislav Plesa
Many important phenomena in chemistry and biology are realized via dynamical features such as multi-stability, oscillations, and chaos. Construction of novel chemical systems with such finely-tuned dynamics is a challenging problem central to the growing field of synthetic biology. In this paper, we address this problem by putting forward a molecular version of a recurrent artificial neural network, which we call a recurrent neural chemical reaction network (RNCRN). We prove that the RNCRN, with sufficiently many auxiliary chemical species and suitable fast reactions, can be systematically trained to achieve any dynamics. This approximation ability is shown to hold independent of the initial conditions for the auxiliary species, making the RNCRN more experimentally feasible. To demonstrate the results, we present a number of relatively simple RNCRNs trained to display a variety of biologically-important dynamical features.
{"title":"Recurrent neural chemical reaction networks that approximate arbitrary dynamics","authors":"Alexander Dack, Benjamin Qureshi, Thomas E. Ouldridge, Tomislav Plesa","doi":"arxiv-2406.03456","DOIUrl":"https://doi.org/arxiv-2406.03456","url":null,"abstract":"Many important phenomena in chemistry and biology are realized via dynamical\u0000features such as multi-stability, oscillations, and chaos. Construction of\u0000novel chemical systems with such finely-tuned dynamics is a challenging problem\u0000central to the growing field of synthetic biology. In this paper, we address\u0000this problem by putting forward a molecular version of a recurrent artificial\u0000neural network, which we call a recurrent neural chemical reaction network\u0000(RNCRN). We prove that the RNCRN, with sufficiently many auxiliary chemical\u0000species and suitable fast reactions, can be systematically trained to achieve\u0000any dynamics. This approximation ability is shown to hold independent of the\u0000initial conditions for the auxiliary species, making the RNCRN more\u0000experimentally feasible. To demonstrate the results, we present a number of\u0000relatively simple RNCRNs trained to display a variety of biologically-important\u0000dynamical features.","PeriodicalId":501325,"journal":{"name":"arXiv - QuanBio - Molecular Networks","volume":"25 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141550855","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We review the construction and evolution of mathematical models of the Arabidopsis circadian clock, structuring the discussion into two distinct historical phases of modeling strategies: extension and reduction. The extension phase explores the bottom-up assembly of regulatory networks introducing as many components and interactions as possible in order to capture the oscillatory nature of the clock. The reduction phase deals with functional decomposition, distilling complex models to their essential dynamical repertoire. Current challenges in this field, including the integration of spatial considerations and environmental influences like light and temperature, are also discussed. The review emphasizes the ongoing need for models that balance molecular detail with practical simplicity.
{"title":"Mathematical models of the Arabidopsis circadian oscillator","authors":"Lucas Henao, Saúl Ares, Pablo Catalán","doi":"arxiv-2405.18006","DOIUrl":"https://doi.org/arxiv-2405.18006","url":null,"abstract":"We review the construction and evolution of mathematical models of the\u0000Arabidopsis circadian clock, structuring the discussion into two distinct\u0000historical phases of modeling strategies: extension and reduction. The\u0000extension phase explores the bottom-up assembly of regulatory networks\u0000introducing as many components and interactions as possible in order to capture\u0000the oscillatory nature of the clock. The reduction phase deals with functional\u0000decomposition, distilling complex models to their essential dynamical\u0000repertoire. Current challenges in this field, including the integration of\u0000spatial considerations and environmental influences like light and temperature,\u0000are also discussed. The review emphasizes the ongoing need for models that\u0000balance molecular detail with practical simplicity.","PeriodicalId":501325,"journal":{"name":"arXiv - QuanBio - Molecular Networks","volume":"18 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-05-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141166702","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Bridget T. McInnes, Jiawei Tang, Darshini Mahendran, Mai H. Nguyen
This paper presents a methodology for enhancing relation extraction from biomedical texts, focusing specifically on chemical-gene interactions. Leveraging the BioBERT model and a multi-layer fully connected network architecture, our approach integrates the ChemProt and DrugProt datasets using a novel merging strategy. Through extensive experimentation, we demonstrate significant performance improvements, particularly in CPR groups shared between the datasets. The findings underscore the importance of dataset merging in augmenting sample counts and improving model accuracy. Moreover, the study highlights the potential of automated information extraction in biomedical research and clinical practice.
{"title":"BioBERT-based Deep Learning and Merged ChemProt-DrugProt for Enhanced Biomedical Relation Extraction","authors":"Bridget T. McInnes, Jiawei Tang, Darshini Mahendran, Mai H. Nguyen","doi":"arxiv-2405.18605","DOIUrl":"https://doi.org/arxiv-2405.18605","url":null,"abstract":"This paper presents a methodology for enhancing relation extraction from\u0000biomedical texts, focusing specifically on chemical-gene interactions.\u0000Leveraging the BioBERT model and a multi-layer fully connected network\u0000architecture, our approach integrates the ChemProt and DrugProt datasets using\u0000a novel merging strategy. Through extensive experimentation, we demonstrate\u0000significant performance improvements, particularly in CPR groups shared between\u0000the datasets. The findings underscore the importance of dataset merging in\u0000augmenting sample counts and improving model accuracy. Moreover, the study\u0000highlights the potential of automated information extraction in biomedical\u0000research and clinical practice.","PeriodicalId":501325,"journal":{"name":"arXiv - QuanBio - Molecular Networks","volume":"32 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-05-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141197110","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：