Dose optimization of an adjuvanted peptide-based personalized neoantigen melanoma vaccine.

The advancements in next-generation sequencing have made it possible to effectively detect somatic mutations, which has led to the development of personalized neoantigen cancer vaccines that are tailored to the unique variants found in a patient's cancer. These vaccines can provide significant clinical benefit by leveraging the patient's immune response to eliminate malignant cells. However, determining the optimal vaccine dose for each patient is a challenge due to the heterogeneity of tumors. To address this challenge, we formulate a mathematical dose optimization problem based on a previous mathematical model that encompasses the immune response cascade produced by the vaccine in a patient. We propose an optimization approach to identify the optimal personalized vaccine doses, considering a fixed vaccination schedule, while simultaneously minimizing the overall number of tumor and activated T cells. To validate our approach, we perform in silico experiments on six real-world clinical trial patients with advanced melanoma. We compare the results of applying an optimal vaccine dose to those of a suboptimal dose (the dose used in the clinical trial and its deviations). Our simulations reveal that an optimal vaccine regimen of higher initial doses and lower final doses may lead to a reduction in tumor size for certain patients. Our mathematical dose optimization offers a promising approach to determining an optimal vaccine dose for each patient and improving clinical outcomes.

Dosage optimization for reducing tumor burden using a phenotype-structured population model with a drug-resistance continuum.

Drug resistance is a significant obstacle to effective cancer treatment. To gain insights into how drug resistance develops, we adopted a concept called fitness landscape and employed a phenotype-structured population model by fitting to a set of experimental data on a drug used for ovarian cancer, olaparib. Our modeling approach allowed us to understand how a drug affects the fitness landscape and track the evolution of a population of cancer cells structured with a spectrum of drug resistance. We also incorporated pharmacokinetic (PK) modeling to identify the optimal dosages of the drug that could lead to long-term tumor reduction. We derived a formula that indicates that maximizing variation in plasma drug concentration over a dosing interval could be important in reducing drug resistance. Our findings suggest that it may be possible to achieve better treatment outcomes with a drug dose lower than the levels recommended by the drug label. Acknowledging the current limitations of our work, we believe that our approach, which combines modeling of both PK and drug resistance evolution, could contribute to a new direction for better designing drug treatment regimens to improve cancer treatment.

Benefit-risk assessment of Covid-19 vaccine, MRNA (MRNA-1273) for males age 18-64 years.

Since the authorization of the Moderna mRNA COVID-19 vaccine, real-world evidence has indicated its effectiveness in preventing COVID-19 cases. However, increased cases of mRNA vaccine-associated myocarditis/pericarditis have been reported, predominantly in young adults and adolescents. The Food and Drug Administration conducted a benefit-risk assessment to inform the review of the Biologics License Application for use of the Moderna vaccine among individuals ages 18 and older. We modeled the benefit-risk per million individuals who receive two complete doses of the vaccine. Benefit endpoints were vaccine-preventable COVID-19 cases, hospitalizations, intensive care unit (ICU) admissions, and deaths. The risk endpoints were vaccine-related myocarditis/pericarditis cases, hospitalizations, ICU admissions, and deaths. The analysis was conducted on the age-stratified male population due to data signals and previous work showing males to be the main risk group. We constructed six scenarios to evaluate the impact of uncertainty associated with pandemic dynamics, vaccine effectiveness (VE) against novel variants, and rates of vaccine-associated myocarditis/pericarditis cases on the model results. For our most likely scenario, we assumed the US COVID-19 incidence was for the week of December 25, 2021, with a VE of 30% against cases and 72% against hospitalization with the Omicron-dominant strain. Our source for estimating vaccine-attributable myocarditis/pericarditis rates was FDA's CBER Biologics Effectiveness and Safety (BEST) System databases. Overall, our results supported the conclusion that the benefits of the vaccine outweigh its risks. Remarkably, we predicted vaccinating one million 18-25 year-old males would prevent 82,484 cases, 4,766 hospitalizations, 1,144 ICU admissions, and 51 deaths due to COVID-19, comparing to 128 vaccine-attributable myocarditis/pericarditis cases, 110 hospitalizations, zero ICU admissions, and zero deaths. Uncertainties in the pandemic trajectory, effectiveness of vaccine against novel variants, and vaccine-attributable myocarditis/pericarditis rate are important limitations of our analysis. Also, the model does not evaluate potential long-term adverse effects due to either COVID-19 or vaccine-attributable myocarditis/pericarditis.

A machine learning approach for identifying variables associated with risk of developing neutralizing antidrug antibodies to factor VIII.

A key unmet need in the management of hemophilia A (HA) is the lack of clinically validated markers that are associated with the development of neutralizing antibodies to Factor VIII (FVIII) (commonly referred to as inhibitors). This study aimed to identify relevant biomarkers for FVIII inhibition using Machine Learning (ML) and Explainable AI (XAI) using the My Life Our Future (MLOF) research repository. The dataset includes biologically relevant variables such as age, race, sex, ethnicity, and the variants in the F8 gene. In addition, we previously carried out Human Leukocyte Antigen Class II (HLA-II) typing on samples obtained from the MLOF repository. Using this information, we derived other patient-specific biologically and genetically important variables. These included identifying the number of foreign FVIII derived peptides, based on the alignment of the endogenous FVIII and infused drug sequences, and the foreign-peptide HLA-II molecule binding affinity calculated using NetMHCIIpan. The data were processed and trained with multiple ML classification models to identify the top performing models. The top performing model was then chosen to apply XAI via SHAP, (SHapley Additive exPlanations) to identify the variables critical for the prediction of FVIII inhibitor development in a hemophilia A patient. Using XAI we provide a robust and ranked identification of variables that could be predictive for developing inhibitors to FVIII drugs in hemophilia A patients. These variables could be validated as biomarkers and used in making clinical decisions and during drug development. The top five variables for predicting inhibitor development based on SHAP values are: (i) the baseline activity of the FVIII protein, (ii) mean affinity of all foreign peptides for HLA DRB 3, 4, & 5 alleles, (iii) mean affinity of all foreign peptides for HLA DRB1 alleles), (iv) the minimum affinity among all foreign peptides for HLA DRB1 alleles, and (v) F8 mutation type.

A Retrospective Review of Center for Biologics Evaluation and Research Advisory Committee Meetings in the Context of the FDA's Benefit-Risk Framework.

The US FDA Center for Biologics Evaluation and Research (CBER) is responsible for the regulation of biologically derived products. FDA has established Advisory Committees (AC) as vehicles to seek external expert advice on scientific and technical matters related to the development and evaluation of products regulated by the agency. We aimed to identify and evaluate common topics discussed in CBER AC meetings during the regulatory decision-making process for biological products and medical devices. We analyzed the content of 119 CBER-led AC meetings between 2009 and 2021 listed on the FDA AC webpage. We reviewed publicly available meeting materials such as briefing documents, summaries, and transcripts. Using a structured review codebook based on FDA benefit-risk guidance, we identified important considerations within the benefit-risk dimensions discussed at the AC meetings: therapeutic context, benefit, risk and risk management, and benefit-risk trade-off, where evidence and uncertainty are critical parts of the FDA benefit-risk framework. Based on a detailed review of 24 topics discussed in 23 selected AC meetings conducted between 2016 and 2021, the two most frequently discussed considerations were "Uncertainty about assessment of the safety profile" and "Uncertainty about assessment of the benefit based on clinical trial data" (16/24 times each) as defined in our codebook. Most of the reviewed meetings discussed Investigational New Drug or Biologics License Applications of products. This review could help sponsors better plan and design studies by contextualizing how the benefit-risk dimensions were embedded in the AC discussions and the considerations that went into the final AC recommendations.

Analysis of tumor-immune functional responses in a mathematical model of neoantigen cancer vaccines.

Cancer neoantigen vaccines have emerged as a promising approach to stimulating the immune system to fight cancer. We propose a simple model including key elements of cancer-immune interactions and conduct a phase plane analysis to understand the immunological mechanisms of cancer neoantigen vaccines. Analytical results are obtained for two widely used functional forms that represent the killing rate of tumor cells by immune cells: the law of mass action (LMA) and the dePillis-Radunskaya Law (LPR). Using the LMA, our results reveal that a slowly growing tumor can escape the immune surveillance and that there is a unique periodic solution. The LPR offers richer dynamics, in which tumor elimination and uncontrolled tumor growth are both present. We show that tumor elimination requires sufficient number of initial activated T cells in relationship to the malignant cells, which lends support to using the neoantigen cancer vaccine as an adjuvant therapy after the primary tumor is surgically removed or treated using radiotherapy. We also derive a sufficient condition for uncontrolled tumor growth under the assumption of the LPR. The juxtaposition of analyses with these two different choices for the killing rate function highlights their importance on model behavior and biological implications, by which we hope to spur further theoretical and experimental work to understand mechanisms underlying different functional forms for the killing rate.

Benefit-risk assessment of COVID-19 vaccine, mRNA (Comirnaty) for age 16-29 years.

Since authorization of the Pfizer-BioNTech COVID-19 Vaccine, mRNA (Comirnaty), real-world evidence has indicated the vaccines are effective in preventing COVID-19 cases and related hospitalizations and deaths. However, increased cases of myocarditis/pericarditis have been reported in the United States associated with vaccination, particularly in adolescents and young adults. FDA conducted a benefit-risk assessment to determine whether the benefits of vaccination outweigh the risks among various age (16-17, 18-24, 25-29) and sex (M/F) subgroups being considered for approved use of the vaccine. We conducted a simulation study with sensitivity analysis of the benefits and risks of the vaccine across possible pandemic scenarios. The model results show benefits outweigh the risks for all scenarios including the high-risk subgroup, males 16-17 years old. Our worst-case scenario used sex and age subgroup-specific incidences for COVID-19 cases (47-98 per million per day) and hospitalizations (1-4 per million per day) which are the US COVID-19 incidences as of July 10, 2021, vaccine efficacy of 70% against COVID-19 cases and 80% against hospitalization, and unlikely, pessimistic, non-zero vaccine-attributable myocarditis death rate. For males 16-17 years old, the model predicts prevented COVID cases, hospitalizations, ICUs, and deaths of 13577, 127, 41, and 1, respectively; while the predicted ranges for excess myocarditis/pericarditis cases, hospitalizations, and deaths attributable to the vaccine are [98-196], [98-196], and 0, respectively, for the worst-case scenario. Considering the different clinical implications of hospitalization due to COVID-19 infection versus vaccine-attributable myocarditis/pericarditis cases, we determine the benefits still outweigh the risks even for this high-risk subgroup. Our results demonstrate that the benefits of the vaccine outweigh its risks for all age and sex subgroups we analyze in this study. Uncertainties exist in this assessment as both benefits and risks of vaccination may change with the continuing evolution of the pandemic.

Mathematical model of a personalized neoantigen cancer vaccine and the human immune system.

Cancer vaccines are an important component of the cancer immunotherapy toolkit enhancing immune response to malignant cells by activating CD4+ and CD8+ T cells. Multiple successful clinical applications of cancer vaccines have shown good safety and efficacy. Despite the notable progress, significant challenges remain in obtaining consistent immune responses across heterogeneous patient populations, as well as various cancers. We present a mechanistic mathematical model describing key interactions of a personalized neoantigen cancer vaccine with an individual patient's immune system. Specifically, the model considers the vaccine concentration of tumor-specific antigen peptides and adjuvant, the patient's major histocompatibility complexes I and II copy numbers, tumor size, T cells, and antigen presenting cells. We parametrized the model using patient-specific data from a clinical study in which individualized cancer vaccines were used to treat six melanoma patients. Model simulations predicted both immune responses, represented by T cell counts, to the vaccine as well as clinical outcome (determined as change of tumor size). This model, although complex, can be used to describe, simulate, and predict the behavior of the human immune system to a personalized cancer vaccine.

A Systematic Review of the Efforts and Hindrances of Modeling and Simulation of CAR T-cell Therapy.

Chimeric antigen receptor (CAR) T-cell therapy is an immunotherapy that has recently become highly instrumental in the fight against life-threatening diseases. A variety of modeling and computational simulation efforts have addressed different aspects of CAR T-cell therapy, including T-cell activation, T- and malignant cell population dynamics, therapeutic cost-effectiveness strategies, and patient survival. In this article, we present a systematic review of those efforts, including mathematical, statistical, and stochastic models employing a wide range of algorithms, from differential equations to machine learning. To the best of our knowledge, this is the first review of all such models studying CAR T-cell therapy. In this review, we provide a detailed summary of the strengths, limitations, methodology, data used, and data gap in currently published models. This information may help in designing and building better models for enhanced prediction and assessment of the benefit-risk balance associated with novel CAR T-cell therapies, as well as with the data need for building such models.

A Quantitative Evaluation of COVID-19 Epidemiological Models.

Quantifying how accurate epidemiological models of COVID-19 forecast the number of future cases and deaths can help frame how to incorporate mathematical models to inform public health decisions. Here we analyze and score the predictive ability of publicly available COVID-19 epidemiological models on the COVID-19 Forecast Hub. Our score uses the posted forecast cumulative distributions to compute the log-likelihood for held-out COVID-19 positive cases and deaths. Scores are updated continuously as new data become available, and model performance is tracked over time. We use model scores to construct ensemble models based on past performance. Our publicly available quantitative framework may aid in improving modeling frameworks, and assist policy makers in selecting modeling paradigms to balance the delicate trade-offs between the economy and public health.

NERDSS: A Nonequilibrium Simulator for Multibody Self-Assembly at the Cellular Scale.

Currently, a significant barrier to building predictive models of cellular self-assembly processes is that molecular models cannot capture minutes-long dynamics that couple distinct components with active processes, whereas reaction-diffusion models cannot capture structures of molecular assembly. Here, we introduce the nonequilibrium reaction-diffusion self-assembly simulator (NERDSS), which addresses this spatiotemporal resolution gap. NERDSS integrates efficient reaction-diffusion algorithms into generalized software that operates on user-defined molecules through diffusion, binding and orientation, unbinding, chemical transformations, and spatial localization. By connecting the fast processes of binding with the slow timescales of large-scale assembly, NERDSS integrates molecular resolution with reversible formation of ordered, multisubunit complexes. NERDSS encodes models using rule-based formatting languages to facilitate model portability, usability, and reproducibility. Applying NERDSS to steps in clathrin-mediated endocytosis, we design multicomponent systems that can form lattices in solution or on the membrane, and we predict how stochastic but localized dephosphorylation of membrane lipids can drive lattice disassembly. The NERDSS simulations reveal the spatial constraints on lattice growth and the role of membrane localization and cooperativity in nucleating assembly. By modeling viral lattice assembly and recapitulating oscillations in protein expression levels for a circadian clock model, we illustrate the adaptability of NERDSS. NERDSS simulates user-defined assembly models that were previously inaccessible to existing software tools, with broad applications to predicting self-assembly in vivo and designing high-yield assemblies in vitro.

An implicit lipid model for efficient reaction-diffusion simulations of protein binding to surfaces of arbitrary topology.

Localization of proteins to a membrane is an essential step in a broad range of biological processes such as signaling, virion formation, and clathrin-mediated endocytosis. The strength and specificity of proteins binding to a membrane depend on the lipid composition. Single-particle reaction-diffusion methods offer a powerful tool for capturing lipid-specific binding to membrane surfaces by treating lipids explicitly as individual diffusible binding sites. However, modeling lipid particle populations is expensive. Here, we present an algorithm for reversible binding of proteins to continuum surfaces with implicit lipids, providing dramatic speed-ups to many body simulations. Our algorithm can be readily integrated into most reaction-diffusion software packages. We characterize changes to kinetics that emerge from explicit vs implicit lipids as well as surface adsorption models, showing excellent agreement between our method and the full explicit lipid model. Compared to models of surface adsorption, which couple together binding affinity and lipid concentration, our implicit lipid model decouples them to provide more flexibility for controlling surface binding properties and lipid inhomogeneity, thus reproducing binding kinetics and equilibria. Crucially, we demonstrate our method's application to membranes of arbitrary curvature and topology, modeled via a subdivision limit surface, again showing excellent agreement with explicit lipid simulations. Unlike adsorption models, our method retains the ability to bind lipids after proteins are localized to the surface (through, e.g., a protein-protein interaction), which can greatly increase the stability of multiprotein complexes on the surface. Our method will enable efficient cell-scale simulations involving proteins localizing to realistic membrane models, which is a critical step for predictive modeling and quantification of in vitro and in vivo dynamics.

TCPro: an In Silico Risk Assessment Tool for Biotherapeutic Protein Immunogenicity.

Most immune responses to biotherapeutic proteins involve the development of anti-drug antibodies (ADAs). New drugs must undergo immunogenicity assessments to identify potential risks at early stages in the drug development process. This immune response is T cell-dependent. Ex vivo assays that monitor T cell proliferation often are used to assess immunogenicity risk. Such assays can be expensive and time-consuming to carry out. Furthermore, T cell proliferation requires presentation of the immunogenic epitope by major histocompatibility complex class II (MHCII) proteins on antigen-presenting cells. The MHC proteins are the most diverse in the human genome. Thus, obtaining cells from subjects that reflect the distribution of the different MHCII proteins in the human population can be challenging. The allelic frequencies of MHCII proteins differ among subpopulations, and understanding the potential immunogenicity risks would thus require generation of datasets for specific subpopulations involving complex subject recruitment. We developed TCPro, a computational tool that predicts the temporal dynamics of T cell counts in common ex vivo assays for drug immunogenicity. Using TCPro, we can test virtual pools of subjects based on MHCII frequencies and estimate immunogenicity risks for different populations. It also provides rapid and inexpensive initial screens for new biotherapeutics and can be used to determine the potential immunogenicity risk of new sequences introduced while bioengineering proteins. We validated TCPro using an experimental immunogenicity dataset, making predictions on the population-based immunogenicity risk of 15 protein-based biotherapeutics. Immunogenicity rankings generated using TCPro are consistent with the reported clinical experience with these therapeutics.

Predictive model for Zika virus RNA minipool nucleic acid testing in outbreak scenarios.

BACKGROUND: Zika virus (ZIKV), a mosquito-borne flavivirus, causes asymptomatic infections in blood donors and can be transmitted by transfusion. During the 2016 US outbreak, universal individual-donation nucleic acid testing (ID-NAT) was used to screen the blood supply for ZIKV. Testing pooled samples from multiple donations with minipool (MP)-NAT is less sensitive than ID-NAT, which raised questions about its utility in ZIKV outbreaks. STUDY DESIGN AND METHODS: A mathematical model and computer simulation determined the risk of missing ID-NAT-reactive and immunoglobulin (Ig) M-negative donations in a ZIKV outbreak if MP-NAT is used initially instead of ID-NAT. The model calculated the time required for ZIKV RNA to replicate to a concentration detectable by testing donations individually or in pools of 6 (MP6) or 16 (MP16). A computer simulation then randomly selected infection times to determine the probability of detection by the candidate tests. RESULTS: The probability of detecting the first ID-NAT-reactive unit in an outbreak is 92% (2.5th-97.5th percentile, 79%-99%) by MP6 and 85% (2.5th-97.5th percentile, 67%-99%) by MP16. When one donation is detected by MP-NAT, the model predicts that the chance of having missed one or more ID-NAT-reactive donations is 8% to 15%. The probability of missing a unit by MP-NAT is constant over the course of the outbreak (8% by MP6, 15% by MP16). CONCLUSION: The model predicts that the probability that a candidate MP-NAT will detect the first ID-NAT-reactive unit in a ZIKV outbreak is 85% to 92% and remains constant over time.

SampPick: Selection of a Cohort of Subjects Matching a Population HLA Distribution.

Immune responses to therapeutic proteins and peptides can adversely affect their safety and efficacy; consequently, immunogenicity risk-assessments are part of the development, licensure and clinical use of these products. In most cases the development of anti-drug antibodies is mediated by T cells which requires antigen presentation by Major Histocompatibility Complex Class II (MHCII) molecules (also called Human Leucocyte Antigen, HLA in humans). Immune responses to many protein therapeutics are thus HLA-restricted and it is important that the distribution of HLA variants used in the immunogenicity assessments provides adequate coverage of the target population. Due to biases inherent to the collection of samples in a blood bank or donor pool, simple random sampling will not achieve a truly representative sample of the population of interest. To help select a donor cohort we introduce SampPick, an implementation of simulated annealing which optimizes cohort selection to closely match the frequency distribution of a target population or subpopulation. With inputs of a target background frequency distribution for a population and a set of available, HLA-typed donors, the algorithm will iteratively create a cohort of donors of a user selected size that will closely match the target population rather than a random sample. In addition to optimizing the HLA types of donor cohorts, the software presented can be used to optimize donor cohorts for any other biallelic or monoallelic trait.

Cytosolic proteins can exploit membrane localization to trigger functional assembly.

Cell division, endocytosis, and viral budding would not function without the localization and assembly of protein complexes on membranes. What is poorly appreciated, however, is that by localizing to membranes, proteins search in a reduced space that effectively drives up concentration. Here we derive an accurate and practical analytical theory to quantify the significance of this dimensionality reduction in regulating protein assembly on membranes. We define a simple metric, an effective equilibrium constant, that allows for quantitative comparison of protein-protein interactions with and without membrane present. To test the importance of membrane localization for driving protein assembly, we collected the protein-protein and protein-lipid affinities, protein and lipid concentrations, and volume-to-surface-area ratios for 46 interactions between 37 membrane-targeting proteins in human and yeast cells. We find that many of the protein-protein interactions between pairs of proteins involved in clathrin-mediated endocytosis in human and yeast cells can experience enormous increases in effective protein-protein affinity (10-1000 fold) due to membrane localization. Localization of binding partners thus triggers robust protein complexation, suggesting that it can play an important role in controlling the timing of endocytic protein coat formation. Our analysis shows that several other proteins involved in membrane remodeling at various organelles have similar potential to exploit localization. The theory highlights the master role of phosphoinositide lipid concentration, the volume-to-surface-area ratio, and the ratio of 3D to 2D equilibrium constants in triggering (or preventing) constitutive assembly on membranes. Our simple model provides a novel quantitative framework for interpreting or designing in vitro experiments of protein complexation influenced by membrane binding.

Collective cancer cell invasion induced by coordinated contractile stresses.

The physical underpinnings of fibrosarcoma cell dissemination from a tumor in a surrounding collagen-rich matrix are poorly understood. Here we show that a tumor spheroid embedded in a 3D collagen matrix exerts large contractile forces on the matrix before invasion. Cell invasion is accompanied by complex spatially and temporally dependent patterns of cell migration within and at the surface of the spheroids that are fundamentally different from migratory patterns of individual fibrosarcoma cells homogeneously distributed in the same type of matrix. Cells display a continuous transition from a round morphology at the spheroid core, to highly aligned elongated morphology at the spheroid periphery, which depends on both ß1-integrin-based cell-matrix adhesion and myosin II/ROCK-based cell contractility. This isotropic-to-anisotropic transition corresponds to a shift in migration, from a slow and unpolarized movement at the core, to a fast, polarized and persistent one at the periphery. Our results also show that the ensuing collective invasion of fibrosarcoma cells is induced by anisotropic contractile stresses exerted on the surrounding matrix.

Theory of bi-molecular association dynamics in 2D for accurate model and experimental parameterization of binding rates.

The dynamics of association between diffusing and reacting molecular species are routinely quantified using simple rate-equation kinetics that assume both well-mixed concentrations of species and a single rate constant for parameterizing the binding rate. In two-dimensions (2D), however, even when systems are well-mixed, the assumption of a single characteristic rate constant for describing association is not generally accurate, due to the properties of diffusional searching in dimensions d ≤ 2. Establishing rigorous bounds for discriminating between 2D reactive systems that will be accurately described by rate equations with a single rate constant, and those that will not, is critical for both modeling and experimentally parameterizing binding reactions restricted to surfaces such as cellular membranes. We show here that in regimes of intrinsic reaction rate (ka) and diffusion (D) parameters ka/D > 0.05, a single rate constant cannot be fit to the dynamics of concentrations of associating species independently of the initial conditions. Instead, a more sophisticated multi-parametric description than rate-equations is necessary to robustly characterize bimolecular reactions from experiment. Our quantitative bounds derive from our new analysis of 2D rate-behavior predicted from Smoluchowski theory. Using a recently developed single particle reaction-diffusion algorithm we extend here to 2D, we are able to test and validate the predictions of Smoluchowski theory and several other theories of reversible reaction dynamics in 2D for the first time. Finally, our results also mean that simulations of reactive systems in 2D using rate equations must be undertaken with caution when reactions have ka/D > 0.05, regardless of the simulation volume. We introduce here a simple formula for an adaptive concentration dependent rate constant for these chemical kinetics simulations which improves on existing formulas to better capture non-equilibrium reaction dynamics from dilute to dense systems.

A mechanochemical model of actin filaments.

In eukaryotic cells, actin filaments are involved in important processes such as motility, division, cell shape regulation, contractility, and mechanosensation. Actin filaments are polymerized chains of monomers, which themselves undergo a range of chemical events such as ATP hydrolysis, polymerization, and depolymerization. When forces are applied to F-actin, in addition to filament mechanical deformations, the applied force must also influence chemical events in the filament. We develop an intermediate-scale model of actin filaments that combines actin chemistry with filament-level deformations. The model is able to compute mechanical responses of F-actin during bending and stretching. The model also describes the interplay between ATP hydrolysis and filament deformations, including possible force-induced chemical state changes of actin monomers in the filament. The model can also be used to model the action of several actin-associated proteins, and for large-scale simulation of F-actin networks. All together, our model shows that mechanics and chemistry must be considered together to understand cytoskeletal dynamics in living cells.