Biology subjects

Hernandez-Vargas, E. A.

Publications and source records attributed to Hernandez-Vargas, E. A..

9 recordsLinked to original sources

bioRxiv2024

Hybrid Neural Differential Equations to Model Unknown Mechanisms and States in Biology

Hybrid neural differential equations (HNDEs) embed neural network components within mechanistic scaffolds, combining the structural interpretability of domain-derived models with the approximation power of neural dynamics. Despite their growing adoption in biology and engineering, neural augmentation can introduce observational degeneracies that compromise mechanistic identifiability and scientific interpretability. In this paper, we develop a theoretical framework for practical preservation of mechanistic identifiability in HNDEs. We formalize bounded neural correction classes and derive Gronwall-type trajectory and observational discrepancy bounds linking neural perturbations to mechanistic parameter ambiguity. We further establish sufficient conditions under which hybrid neural corrections preserve approximate mechanistic parameter recoverability up to explicitly quantifiable tolerances. Empirical likelihood profile analyses on benchmark systems confirm that neural augmentation systematically weakens--but does not eliminate--mechanistic identifiability, revealing a fundamental expressiveness-identifiability trade-off. These results provide theoretical foundations and actionable guidance for deploying HNDEs in scientific intelligent computing.

systems biology↗

bioRxiv2024

Invariant sets for switched systems to tackle bacterial resistance

The limited availability of antibiotics and the need for prompt decision-making present significant challenges for healthcare practitioners. When faced with this situation, practitioners must prioritize their approach based on several key factors. By leveraging the emergent understanding of collateral sensitivity among antibiotic-exposed pathogens, we demonstrate the utility of control invariant sets to predict treatment failure when antibiotic cycling is applied as a therapeutic strategy aiming to eradicate or prevent emergence of multi-drug resistant pathogens. Our results here pave the way for point-of-care diagnostic technologies to identify infections and select appropriate treatments quickly, reducing unnecessary antibiotic use.

bioinformatics↗

bioRxiv2020

Switching Logistic Maps to Design Cycling Approaches Against Antimicrobial Resistance

Antimicrobial resistance is a major threat to global health and food security today. Scheduling cycling therapies by targeting phenotypic states associated to specific mutations can help us to eradicate pathogenic variants in chronic infections. In this paper, we introduce a logistic switching model in order to abstract mutation networks of collateral resistance. We found particular conditions for which unstable zero-equilibrium of the logistic maps can be stabilized through a switching signal. That is, persistent populations can be eradicated through tailored switching regimens. Starting from an optimal-control formulation, the switching policies show their potential in the stabilization of the zero-equilibrium for dynamics governed by logistic maps. However, employing such switching strategies, deserve a specific characterization in terms of limit behaviour. Ultimately, we use evolutionary and control algorithms to find either optimal and sub-optimal switching policies. Simulations results show the applicability of Parrondos Paradox to design cycling therapies against drug resistance.

bioengineering↗

bioRxiv2019

Quantitative modeling of the function of kinetically driven transcriptional riboswitches

We propose a generalized modeling framework for the kinetic mechanisms of transcriptional riboswitches. The formalism accommodates time-dependent transcription rates and changes of metabolite concentration and permits incorporation of variations in transcription rate depending on transcript length. We derive explicit analytical expressions for the fraction of transcripts that determine repression or activation of gene expression, pause site location and its slowing down of transcription for the case of the (2dG)-sensing riboswitch from Mesoplasma florum. Our modeling challenges the current view on the exclusive importance of metabolite binding to transcripts containing only the aptamer domain. Numerical simulations of transcription proceeding in a continuous manner under time-dependent changes of metabolite concentration further suggest that rapid modulations in concentration result in a reduced dynamic range for riboswitch function regardless of transcription rate, while a combination of slow modulations and small transcription rates ensures a wide range of finely tuneable regulatory outcomes.

systems biology↗

bioRxiv2019

Topological data analysis to uncover the shape of immune responses during co-infection

Co-infections by multiple pathogens have important implications in many aspects of health, epidemiology and evolution. However, how to disentangle the contributing factors of the immune response when two infections take place at the same time is largely unexplored. Using data sets of the immune response during influenza-pneumococcal co-infection in mice, we employ here topological data analysis to simplify and visualise high dimensional data sets.\n\nWe identified persistent shapes of the simplicial complexes of the data in the three infection scenarios: single viral infection, single bacterial infection, and co-infection. The immune response was found to be distinct for each of the infection scenarios and we uncovered that the immune response during the co-infection has three phases and two transition points. During the first phase, its dynamics is inherited from its response to the primary (viral) infection. The immune response has an early (few hours post co-infection) and then modulates its response to finally react against the secondary (bacterial) infection. Between 18 to 26 hours post co-infection the nature of the immune response changes again and does no longer resembles either of the single infection scenarios.\n\nAuthor summaryThe mapper algorithm is a topological data analysis technique used for the qualitative analysis, simplification and visualisation of high dimensional data sets. It generates a low-dimensional image that captures topological and geometric information of the data set in high dimensional space, which can highlight groups of data points of interest and can guide further analysis and quantification.\n\nTo understand how the immune system evolves during the co-infection between viruses and bacteria, and the role of specific cytokines as contributing factors for these severe infections, we use Topological Data Analysis (TDA) along with an extensive semi-unsupervised parameter value grid search, and k-nearest neighbour analysis.\n\nWe find persistent shapes of the data in the three infection scenarios, single viral and bacterial infections and co-infection. The immune response is shown to be distinct for each of the infections scenarios and we uncover that the immune response during the co-infection has three phases and two transition points, a previously unknown property regarding the dynamics of the immune response during co-infection.

immunology↗

bioRxiv2019

Focused natural product elucidation by prioritizing high-throughput metabolomic studies with machine learning

Bacteria of the genera Photorhabdus and Xenorhabdus produce a plethora of natural products to support their similar symbiotic lifecycles. For many of these compounds, the specific bioactivities are unknown. One common challenge in natural product research when trying to prioritize research efforts is the rediscovery of identical (or highly similar) compounds from different strains. Linking genome sequence to metabolite production can help in overcoming this problem. However, sequences are typically not available for entire collections of organisms. Here we perform a comprehensive metabolic screening using HPLC-MS data associated with a 114-strain collection (58 Photorhabdus and 56 Xenorhabdus) from across Thailand and explore the metabolic variation among the strains, matched with several abiotic factors. We utilize machine learning in order to rank the importance of individual metabolites in determining all given metadata. With this approach, we were able to prioritize metabolites in the context of natural product investigations, leading to the identification of previously unknown compounds. The top three highest-ranking features were associated with Xenorhabdus and attributed to the same chemical entity, cyclo(tetrahydroxybutyrate). This work addresses the need for prioritization in high-throughput metabolomic studies and demonstrates the viability of such an approach in future research.

microbiology↗

bioRxiv2018

The trichotomy of pneumococcal infection outcomes

The successful elimination of bacteria such as Streptococcus pneumoniae from a host involves the coordination between different parts of the immune system. Previous studies have explored the effects of the initial pneumococcal load (bacterial dose) on different representations of innate immunity, finding that pathogenic outcomes can vary with the size of the bacterial dose. However, others yield support to the notion of dose-independent factors contributing to bacterial clearance. In this paper, we seek to provide a deeper understanding of the immune responses associated to the pneumococcus. To this end, we formulate a model that realizes an abstraction of the innate-regulatory immune host response. Stability and bifurcation analyses of the model reveal the following trichotomy of pneumococcal outcomes determined by the bifurcation parameters: (i) dose-independent clearance; (ii) dose-independent persistence; and (iii) dose-limited clearance. Bistability, where the bacteria-free equilibrium co-stabilizes with the most substantial steady-state bacterial load is the specific result behind dose-limited clearance. The trichotomy of pneumococcal outcomes here described integrates all previously observed bacterial fates into a unified framework.

microbiology↗

bioRxiv2017

Multiscale Model Within-host and Between-host for Viral Infectious Diseases

Multiscale models possess the potential to uncover new insights into infectious diseases. Here, a rigorous stability analysis of a multiscale model within-host and between-host is presented. The within-host model describes virus replication and the respective immune response while disease transmission is represented by a simple susceptible-infected (SI) model.\n\nThe bridge of within-to between-host is by considering transmission as a function of the viral load of the within-host level. Consequently, stability and bifurcation analyses were developed coupling the two basic reproduction numbers [Formula] and [Formula]for the within- and the between-host subsystems, respectively. Local stability results for each subsystem, such as a unique stable equilibrium point, recapitulate classical approaches to infection and epidemic control.\n\nUsing a Lyapunov function, global stability of the between-host system was obtained. A main result was the derivation of the [Formula] as a general increasing function of [Formula]. Numerical analyses reveal that a Michaelis-Menten form based on the virus is more likely to recapitulate the behavior between the scales than a form directly proportional to the virus. Our work contributes basic understandings of the two models and casts light on the potential effects of the coupling function on linking the two scales.

epidemiology↗

bioRxiv2017

Multiscale Modeling To Explore Ebola Vaccination Strategies

BackgroundTransmission in epidemics of infectious diseases is characterized by a high level of subject-specific elements. These include heterogeneous infection conditions, time-dependent transmission potential, and age-dependent contact structure. These insights are often lost in epidemic models using population data. Here we submit an approach that can capture these details, paving the way for studying epidemics in a more mechanistic and realistic way.\n\nMethodsUsing experimental data, we formulated mathematical models of a pathogen infection dynamics from which we can simulate its transmission potential mechanistically. The models were then embedded in our implement of an age-specific contact network structure that allows to express all elements relevant to the transmission process. This approach is illustrated here with an example of Ebola virus (EBOV).\n\nResultsThe results showed that within-host infection dynamics can capture EBOVs transmission parameters as good as approaches using population data. Population age-structure, contact distribution and patterns can also be captured with our network generating algorithm. This framework opens vast opportunities for the investigations of each element involved in the epidemic process. Here, estimating EBOVs reproduction number revealed a heterogeneous pattern among age-groups, prompting questions on current estimates which are not adjusted for this factor. Assessments of mass vaccination strategies showed that a time window from five months before to one week after the start of an epidemic appeared to be effective. Noticeably, compared to a non-intervention scenario, a low vaccination coverage of 33% could reduce number of cases by ten to hundred times as well as lessen the case-fatality rate.\n\nConclusionsThis is the first effort coupling directly within-host infection model into an age-structured epidemic network model, adding more realistic elements in simulating epidemic processes. Experimental data at the within-host infection are shown able to capture upfront key parameters of a pathogen; the applications of this approach will give us more time to prepare for potential epidemics. Population of interest in epidemic assessments could be modeled with an age-specific contact network without exhaustive amount of data. Further assessments and adaptations for different pathogens and scenarios are underway to explore multilevel aspects in infectious diseases epidemics.

epidemiology↗