The European Physical Journal E

Colorectal cancer (CRC) is the second most common cause of cancer-related mortality. At the molecular level, CRC is associated with genetic mutations and epigenetic modifications that dysregulate various signaling networks. From the biophysical viewpoint, invasive and metastatic cell migration need to be empowered by mechanical forces. In this study, we analyze the dynamic deformation of patient-derived CRC organoids in Fourier space and demonstrate how organoids with protooncogene BRAF mutation exhibit deformation phenotypes at an early stage. The organoids with BRAFmut have significantly lower elasticity and higher viscosity than those with BRAFWT, which mathematically indicated as the weakening of cell-cell adhesion. Immunohistochemical images, qRT-PCR, and TCGA data analysis confirm the downregulation of E-cadherin (CDH1) in BRAFmut organoids as well as in BRAFmut CRC, suggesting that the decrease in cell-cell adhesion in BRAFmut CRC facilitates invasive and metastatic migration. Notably, the recovery of CDH1 expression by pharmacological inhibition of DNA methylation can quantitatively be detected as the change in mechanical properties, suggesting that the complementary combination of dynamic phenotyping, mathematical modelling, and molecular-level analyses has a potential to unravel the mechanistic causality of the critical gene mutation and CRCs prognosis and the response to therapeutic interventions.

Understanding how cells migrate in response to external cues has important implications for biology, medicine, and bioengineering. Chemical, mechanical, and electrical signals are the primary drivers of directed cell migration, and each has been extensively studied over the past decades. Among them, chemical cues were the first to be investigated and remain the most widely studied due to their undeniable role in in vivo guidance. Mechanical signals--particularly substrate stiffness gradients--have gained prominence for their ubiquity across cell types and their potential to direct migration. More recently, growing evidence suggests that electrotaxis offers a highly precise and programmable means to guide cell movement. Despite this, these cues are often studied in isolation, whereas in vivo they typically coexist and interact. Using wellestablished biophysical models, we investigate how mechanical and electrical signals cooperate and how they can be engineered to compete for control over cell migration. We demonstrate that an electric field can override and even reverse durotaxis, with outcomes that depend strongly on the specific cell type. To address this large variability in controlling cell migration, we propose particular steps toward further exploration. To support such future research, we provide a freely available platform for predicting electro-mechanical interactions in cell migration, based on a given cells sensing and signaling characteristics, which could tailor the mechanical and electrical signals that arise naturally during organ development, cancer invasion, or tissue regeneration.

Methods to visualize and quantify the molecular responses of cells to local forces exerted at adhesions are crucial to elucidate how physical forces control cellular behavior. Of the many proteins involved in focal adhesions, vinculin plays a key role in mediating force-sensitive processes. Here, we combined optical tweezers and Forster resonance energy transfer (FRET) microscopy to measure the intensity and FRET efficiency of the vinculin tension sensor, VinTS, in response to a force. Fibroblasts expressing VinTS formed adhesions on fibronectin-coated, 3m-diameter, polystyrene beads. As the beads were displaced by the cell, we applied an optical trap to counteract this movement and increase the traction force required by the cell to maintain the bead displacement. The optical trap stiffness varied from zero (no laser) up to 0.26 pN/nm. In this range, the median bead displacement after 5 min was ~200nm in all trapping conditions inducing counteracting forces in the 10-100pN range. To maintain this displacement, vinculin recruitment increased (up to 35% in relative intensity at high stiffness) while tension increased but more moderately (1-2% decrease in absolute FRET efficiency). For higher trap stiffness, the main response was an increase in vinculin recruitment, while the tension did not increase significantly. The increase in vinculin intensity was correlated with the decrease in FRET efficiency at 0.26 pN/nm but not at lower stiffness. Thus, the presence of the high stiffness optical trap over 5 min appears to induce a positive correlation between vinculin recruitment and vinculin tension. In a few instances, vinculin puncta migrated a few microns away from the bead exceeding the bead movement speed while experiencing an increase in both vinculin intensity and tension. Taken together, the results suggest that combining an optical trap with vinculin tension measurements uncovers novel vinculin dynamics in the presence of a force.

Understanding how cells migrate through confined environments is crucial for elucidating fundamental biological processes, including cancer invasion, immune surveillance, and tissue morphogenesis. The nucleus, as the largest and stiffest cellular organelle, often limits cellular deformability, making it a key factor in migration through narrow pores or highly constrained spaces. In this work, we introduce a geometric surface partial differential equation (GS-PDE) model in which the cell plasma membrane and nuclear envelope are described as evolving energetic closed surfaces governed by force-balance equations. We replicate the results of a biophysical experiment, where a microfluidic device is used to impose compressive stresses on cells by driving them through narrow microchannels under a controlled pressure gradient. The model is validated by reproducing cell entry into the microchannels. A parametric sensitivity analysis highlights the dominant influence of specific parameters, whose accurate estimation is essential for faithfully capturing the experimental setup. We found that surface tension and confinement geometry emerge as key determinants of translocation efficiency. Although tailored to this specific setup for validation purposes, the framework is sufficiently general to be applied to a broad range of cell mechanics scenarios, providing a robust and flexible tool for investigating the interplay between cell mechanics and confinement. It also offers a solid foundation for future extensions integrating more complex biochemical processes such as active confined migration.

It is an essential element of mechanobiology to measure the forces of biological cells. In microparticle traction force microscopy, they are inferred from the deformation of elastic microparticles. Two complementary variants have been introduced before: the volume method, which reconstructs surface stresses from the displacements of fiducial markers embedded inside the particles, and the surface method, which infers stresses directly from the deformation of the particle surface. However, a systematic comparison of the two methods has been lacking. Here, we quantitatively compare both approaches using simulated traction fields representing biologically relevant loading scenarios. We find that the surface method consistently reconstructs traction profiles with substantially lower errors than the volume method, which suffers from displacement tracking and stress calculation at the surface. At high noise levels, however, the performance gap becomes smaller. To compare the performance of the two methods in a realistic experimental setting, we developed DNA-based hydrogel microparticles equipped with both fluorescent surface labels and embedded fluorescent nanoparticles, enabling the direct comparison of the two methods within the same system. Compression experiments produced traction profiles consistent with Hertzian contact mechanics and confirmed the trends observed in the simulations. While our computational workflow establishes a framework to apply both methods, our experimental workflow establishes DNA microparticles as versatile and biocompatible probes for measuring cellular forces.

1Nuclear envelope stretch and rupture are common to cell spreading and migration in a variety of microenvironments, leading to marked changes in nucleocytoplasmic transport. Predicting cell response to different mechanochemical cues that are transmitted to the nucleus remains an open problem in the field of mechanomedicine. We developed a predictive modeling framework to examine how nuclear deformation on substrates with different nanotopographies influences nucleocytoplasmic transport and rearrangement of the nuclear lamina. Using the finite element method, we simulated nuclear compression by the perinuclear actin cap on substrates with arrays of nanopillars, modeling the nuclear envelope as a nonlinear elastic structure and coupling deformations to a biochemical model of lamin remodeling and nucleocytoplasmic transport. These simulations predicted regions of high nuclear envelope stretch adjacent to cell-nanopillar contacts, leading to maximized nuclear envelope tension on small nanopillars spaced by 4-5 microns. We then considered the effects on nuclear transport of YAP and TAZ and found that increased nuclear compression led to YAP/TAZ nuclear localization in agreement with previous experiments. Furthermore, the simulated force load per lamin was maximized on nanopillar substrates with high nuclear stretch. The magnitude of this load was modulated by the rate of actin cap assembly and the overall expression level of lamin A/C - decreasing lamin content in the nuclear envelope led to a higher likelihood of rupture. We validated this prediction in subsequent experiments with lamin-depleted U2OS cells, establishing the central importance of lamin transport and microenvironment nanotopography to nuclear mechanotransduction. 2 SignificanceCell nuclei commonly experience large strains, but existing computational models do not explain the coupling between such deformations and molecular transport. Here, we present a modeling framework that includes the mechanics of nuclear deformations and the reaction-transport of molecules within the cytoplasm, nuclear envelope, and nuclear interior. As a well-controlled setup for comparing experiments and simulations, we consider nuclear indentations exhibited by cells on nanopillar substrates. Our simulations recapitulate measurements of nuclear YAP/TAZ localization from the literature and predict that low-lamin cells experience higher force loads at the nuclear envelope. We validate this prediction experimentally, showing that lamin-depleted cells are more likely to exhibit nuclear rupture. Overall, our framework presents opportunities to predict nuclear mechanoadaptation to different microenvironments.

Nucleus shape is a sensitive indicator of cell state, influenced by numerous bio-chemical and physiological factors. While prior work has cataloged how perturbations alter nucleus morphology, we address the inverse: inferring underlying molecular changes from nucleus shape alone. We previously developed a mechanical model yielding two nondimensional parameters: flatness index and scale factor, which are surrogate measures for cortical actin tension and nuclear envelope compliance respectively. In this study, we apply these parameters to investigate the dynamics in cellular mechanics during confined migration. We fabricated polydimethylsiloxane (PDMS) microchannels with widths of 3 {micro}m (high confinement) and 10 {micro}m (low confinement) and tracked cells migrating through them. We captured high-frequency 3D nucleus shapes via double fluorescence exclusion microscopy and custom image analysis. Fitting the model and estimating flatness index and scale factor to time-resolved shapes revealed dynamic regulation in 3 {micro}m channels: actin tension decreased and nucleus compliance increased immediately before nucleus entry into the constriction, with rapid restoration to baseline upon exit. No such changes occurred in 10 {micro}m channels, indicating active, confinement-dependent cytoskeletal adaptation. Immunostaining for YAP and lamin-A,C confirmed these model inferences. Our results uncover mechanostasis, active mechanical homeostasis, during confined migration and establish the combination of double fluorescence exclusion microscopy and nondimensional nucleus shape parameters as a powerful, non-invasive tool for single-cell mechanobiology studies.

Fluorescence recovery after photobleaching (FRAP) is widely used to characterize diffusion in cells, but quantitative interpretation of the data in small prokaryotes requires explicitly accounting for cell geometry. While this has been successfully achieved for spherical and rod-shaped bacteria, analytical approaches developed in these cases are not directly applicable to cells with more complex morphologies. Here, we explore the application of FRAP to helical bacteria using simulations. We show that half-compartment FRAP experiments, where one-half of the cell is photobleached, provide a robust means of characterizing fast protein diffusion. To help with the practical implementation of this technique, we established the relationship between the diffusion coefficient and characteristic fluorescence recovery time as a function of cell length and helical parameters, and for two different ways of estimating the recovery time. As a first application, we report measurements of the diffusion coefficient of the fluorescent protein, mNeonGreen, in the helical bacterium Paramagnetospirillum magneticum AMB-1. We find it to be D = 4.9 {+/-} 2.2 {micro}m2 s-1 in isosmotic conditions, not significantly different from the value measured in Escherichia coli. Although developed for helical bacteria, including spirilla, spirochetes, and vibrios, our framework can readily be extended to cells or compartments with other geometries.

Molecular dynamics (MD) simulations have yielded important insights into ion conduction in potassium channels, but quantitative comparison with electrophysiological experiments remains challenging. Due to their high computational cost, MD simulations are typically performed at membrane potentials well above physiological values, and at only a limited number of voltages. Since current-voltage relationships are not necessarily linear, this limits direct comparison between simulations and experiments. Here, we introduce a method to estimate the current-voltage characteristics of ion channels from Markov state models (MSMs) constructed from MD simulations performed at only a few membrane potentials. Time-discrete MSMs of ion conduction are converted into continuous-time rate matrices, whose voltage dependence is modelled using a rate theory formulation with free energy barriers depending on membrane potential. This approach enables the prediction of channel currents over a wide voltage range without additional simulations. We validated the method using MD simulations of the potassium channels KcsA and MthK. In both cases, the currents predicted at low membrane potentials are in good agreement with those obtained directly from MD simulations, demonstrating the robustness and efficiency of the approach.

As we know, bacterial motion navigates complex environments in natural settings, providing a basis for scientific study to understand their dynamics. Hydrodynamics drive wall alignment and accumulation, but research remains unclear about the extent to which confinement alone, without hydrodynamics, modulates bacterial dynamics. Therefore, we develop a 2D model to study the effects of isolated steric and hydrodynamic forces on bacterial motion in 10- and 50-m microchannels. We used bacteria to model self-propelled rigid rods with run-and-tumble motion, and then compared dry systems (purely steric wall effects) with wet systems (hydrodynamic effects). We found that bacterial speeds and their orientation are independent of channel dimensions in dry systems. In wet systems, we observed strong wall-hugging and alignment due to wall-induced hydrodynamic interactions, enhanced residence times, and a slight increase in the observable effective speed in 10 m microchannels compared to 50 m channels (where bacteria-maintained bulk-like dynamics). Our study thus emphasises when confinement affects bacterial motion solely due to pure dry geometry, and when hydrodynamics play an essential role. This study provides a template for microfluidics-based experimental prediction of bacterial dynamics and could be applied to an analysis of antibiotic resistance. Significance StatementMicrobes rely on self-propelled motion to navigate complex environmental systems and survive and persist. In such an environment, physical interactions with surrounding boundaries play a key role in how bacteria move, orient, or settle in complex spaces. Although many studies show that hydrodynamic forces near walls influence how bacteria swim in liquid, there is still confusion about whether confinement, with or without hydrodynamic effects, can change the complete bacterial pattern. Our 2D model shows that confinement (purely steric wall effects, dry limit) alone does not alter bacterial motion patterns. It is the hydrodynamics (wet limit) that drives bacteria to orient toward walls, remain near boundaries, and direct motion in narrow channels. This work clarifies when confinement and hydrodynamics actually affect bacterias motion and provides a practical way to understand bacterial behaviour in biological systems, including in microfluidic applications and studies of antibiotic resistance strategies.

We present a fast, unconditionally energy-stable numerical scheme for simulating vesicle deformation under osmotic pressure using a phase-field approach. The model couples an Allen-Cahn equation for the biomembrane interface with a variable-mobility Cahn-Hilliard equation governing mass exchange across the membrane. Classical approaches, including nonlinear multigrid and Multiple Scalar Auxiliary Variable (MSAV) methods, require iterative solution of variable-coefficient systems at each time step, resulting in substantial computational cost. We introduce a constant-coefficient MSAV (CC-MSAV) scheme that incorporates stabilization directly into the Cahn-Hilliard evolution equation rather than the chemical potential. This reformulation yields fully decoupled constant-coefficient elliptic problems solvable via fast discrete cosine transform (DCT), eliminating iterative solvers entirely. The method achieves O(N2 log N) complexity per time step while preserving unconditional energy stability and discrete mass conservation. Numerical experiments verify second-order temporal and spatial accuracy, mass conservation to relative errors below 5 x 10-11, and close agreement with nonlinear multigrid benchmarks. On grids with N [≥] 2048, CC-MSAV achieves 6-15x overall speedup compared to classical MSAV with optimized preconditioning, while the dominant Cahn-Hilliard subsystem is accelerated by up to two orders of magnitude. These efficiency gains, achieved without sacrificing accuracy, make CC-MSAV particularly well-suited for large-scale simulations of vesicle dynamics.

Elastic and viscoelastic properties of extracellular matrices (ECM) are known to regulate cellular behavior and mechanosensation differently, with implications for morphogenesis, wound healing, and pathophysiology. Most in vitro cellular processes, including cell migration, are studied on linear-elastic substrates to mimic extracellular matrices. However, most tissues are viscoelastic and display a loss modulus (G) that may be 10-20% of their storage modulus (G) under biophysically relevant conditions. Recent research has shown that cells can distinguish between elastic and viscoelastic ECM, leading to alterations in their cellular morphology, migration rates, and contractility. Here, we present a protocol for creating PAH-based model ECMs that enables the fabrication of viscoelastic substrates with storage moduli similar to those of their elastic counterparts. To explore how G influences epithelial cell mechanobiology, we fabricated tunable viscoelastic model ECMs with G of 3 kPa, 8 kPa, and 12 kPa, and for each, independently tuned G values to approximately 300 Pa, 500 Pa, and 700 Pa, respectively. We found that A549 cells cultured on stiff elastic model ECMs migrated [~]30% slower and formed larger focal adhesions compared to their viscoelastic counterparts. Conversely, A549 cells on intermediate viscoelastic model ECMs exhibited a [~]54% reduction in migration speed, with no significant difference in focal adhesion size relative to their elastic counterparts. These findings highlight the complex interplay between substrate (ECM) elastic and viscoelastic properties in regulating epithelial cell mechanobiology and emphasize the importance of time-dependent matrix mechanics in governing epithelial responses.

Despite much of the literature perceiving fish schooling as an organized system with a focus on fixed formations for theoretical analyses, experimental observations suggest that frequent positional rearrangement commonly occurs. Previous studies have also demonstrated that fish schools reduce locomotor costs relative to individuals swimming alone. This introduces an intriguing dichotomy. How can individual fish within schools exhibit dynamic interactions while also saving energy? We hypothesize that schooling dynamics are the result of positional and kinematic modulation of individuals responding to fluid dynamic stimuli from the movement of neighbouring individuals. We propose a two-tier approach to studying kinematic modulation within fish schools. First, quantification of the variation of individual movement in a school relative to that of a solitary individual uses an analytical pipeline combining artificial-intelligence-enabled tracking and video processing. Second, the study of kinematic modulation in response to hydrodynamic stimuli uses a mechanical flapping mechanism coupled with an enclosure to control fish position. We discovered that fish in schools exhibit higher levels of positional and kinematic modulation than individuals swimming alone. Fish swimming in enclosures can robustly respond to fluid stimuli from either a simple robotic fish or other fish located in proximity. This two-tier approach allows high-resolution analysis of positional and kinematic modulation within fish schools and their impacts on energy conservation resulting from collective movement.

Background/ObjectivesThe unprecedented structural and binding data for antibodies to the SARS-COV2 virus taken together with the mutations for the spike protein allows for a broad simulation study of antibody-spike protein binding. This provides an understanding of the co-evolution of human immunity and viral immunity escape. MethodsWe utilized the YASARA molecular dynamics program to generate initial antibody-spike structures and simulate to equilibration for six SARS-COV2 variants and 10 different antibodies sampling two different binding regions to the receptor binding domain of the spike (especially for the Class I antibodies in the same part of the spike which attaches to the ACE2 receptor protein) and one to the N-terminal of the spike. Starting structures for antibody binding to variant spike proteins are perturbatively achieved through point mutations and insertions/deletions in the YASARA program. We employed YASARA to measure interfacial hydrogen bound counts between antibodies and variant spike proteins, and the HawkDock MMGBSA program to characterize trends in binding energies with mutation for four of the antibodies. We utilized the VMD program to analyze the time course of hydrogen bond populations. ResultsAs seen in previous studies, interfacial hydrogen bond counts serve as an excellent proxy for binding energies without the large systematic error inherent in the latter. We find that there is generally a decline in antibody binding strength, as measured by interfacial hydrogen bond counts, with viral evolution, but that a modest re-entrance of binding strength is present for most antibodies studied. Generically, the antibody heavy chain binds more strongly to the spike protein, through for approximately half the antibodies the light chain binding strength converges to the heavy chain strength with viral evolution. ConclusionsThe key conclusion is that the identified re-entrant immunity, speculatively arising from a balancing of maintenance of ACE2-spike binding while escaping antibodies through mutation, allows for some maintenance and even strengthening of immunity for later viral strains from early infection or vaccination.

The actin cytoskeleton is an inherently disordered active system. the actomyosin cortex and reconstituted actomyosin systems are globally disordered, yet undergo transitions between distinct disordered states as parameters like motor and crosslinker concentration and filament length and rigidity change. In cells these changes are related to genetic mutations or differences in cell state and dictate fundamental biological processes. However, we dont have well established methods to detect and classify differences in disordered polymer networks. Image-based morphology techniques provide a non-invasive, high-throughput method of extracting information about a system. In this work we simulate biopolymer networks under varying conditions and develop and use morphological descriptors to construct trajectories in morphospace. Using statistical analysis we find that morphological descriptors are able to distinguish between different trajectories of the system, including differences not apparent to the eye. However, no single descriptor alone is able to capture all the differences in the simulated trajectories. Nematic order parameters typically perform the worst for our simulations while curvature and texture descriptors can collectively distinguish between dynamic trajectories. This work helps develop quantification of cytoskeleton dynamics for classification and data-driven modeling.

The unfolding or melting temperature (TM) is a central quantity to characterize the stability of proteins and other biopolymers. The accurate prediction of protein melting temperatures by molecular mechanics force field simulations is highly desirable for many biophysical and biotechnological applications. Since the time scales for protein (un-)folding are hardly accessible in conventional MD (cMD) simulations, enhanced sampling techniques such as Temperature Replica Exchange Molecular Dynamics (TREMD) are typically employed. However, TREMD simulations are computationally very demanding especially if large temperature ranges need to be covered. Additionally, if the TM is initially unknown, setting up TREMD simulations is often challenging. To find the optimal initial conditions for such simulations, we describe their performance based on a theoretical model, which we validate on a minimalistic Markov Chain Monte Carlo (MCMC) simulation setup. In an effort to reduce the computational demand, we have investigated the possibility to use small sets of TREMD temperature ladders placed iteratively in the vicinity of a TM estimate. Different TREMD setups were extensively tested on the fast-folding protein Chignolin. We found that appropriate starting conformations lead to significantly faster convergence. Furthermore, we found that, in practice, combining multiple small temperature ladders can be advantageous in comparison to one single temperature ladder. Based on our findings, we formulate practical recommendations on how to set up TREMD for protein melting with optimal efficiency.

Transition metal based compounds are promising therapeutic agents, particularly in cancer treatment. However, predicting their binding sites remains a major challenge. In this work, we investigate the applicability of two tools, Metal3D and Metal1D, for this purpose. Although originally trained to predict zinc ion binding sites only, both predictors successfully identify several experimentally observed binding sites for transition metal complexes directly from apo protein structures. At the same time, we highlight current limitations, such as the sensitivity to side-chain conformations, and discuss possible strategies for improvement. This work provides a first step toward establishing a robust computational pipeline in which rapid and low-cost predictors are able to identify putative hotspots for transition metal binding, which can then be refined using more accurate but computationally demanding methods. Author summaryTransition metals play a crucial role as therapeutic agents, especially in cancer therapy. However, the prediction of their binding site locations is challenging, as accurate computational methods often require time-consuming simulations, making them impractical when many possible binding sites must be explored. In this work, we explored the capability of two binding site predictors, originally developed to locate metal ions in proteins, to identify binding sites for more complex covalently-bound transition metal based agents. We found that these tools can often identify the experimentally-known binding regions, even when starting from the apo structure, in which the protein does not already contain the metal compound. At the same time, our results show clear limitations in more challenging cases, particularly when the binding involves only a single amino acid or when the binding site undergoes major structural rearrangements upon binding. Overall, our study shows that fast predictors can provide valuable early insights in the investigation of the binding sites of covalently-bound transition metal based compounds. When combined with more accurate simulation techniques, they can help focus computational efforts and ultimately support the rational design of transition metal based drugs.

Single-particle tracking (SPT) is a tool of growing importance which enables biologists to better understand the dynamics of protein interactions at the single-molecule level and in vivo. However, the stochastic nature of the motion of single molecules, the wide variety of types of motion that they can experience and complex transition kinetics between the different states of motion are challenging factors that complicate the interpretation of SPT data. This article presents and benchmarks the tool ExaTrack. Like previous tools, it can handle particles moving in one or multiple states of motion with transitions between states. Its unique feature is that it can simultaneously handle a wide range of complex types of motion such as diffusive motion, directed motion and confined motion while also managing a variety of transition kinetics such as memoryless first-order transitions or more complex time-dependent state transitions. This manuscript focuses on the benchmarking of ExaTrack on simulated data.

Kinase family proteins constitute the second largest protein class targeted in drug development efforts, most prominently to treat cancer, but also several other diseases associated with kinase dysfunction. In this work we focus on type II kinase inhibitors which bind to the "classical" inactive conformation of the protein kinase catalytic domain where the DFG motif has a "DFG-out" orientation and the activation loop is folded. Many Tyrosine kinases (TKs) exhibit strong binding affinity with a wide spectrum of type II inhibitors while serine/threonine kinases (STKs) often bind more weakly. Recent work suggests this difference is largely due to differences in the folded to extended conformational equilibrium of the activation loop between TKs vs. STKs. The binding affinity of a type II inhibitor to its kinase target can be decomposed into a sum of two contributions: (1) the free energy cost to reorganize the protein from the active to inactive state, and (2) the binding affinity of the type II inhibitor to the inactive kinase conformation. In previous work we used a Potts statistical energy potential based on sequence co-variation to thread sequences over ensembles of active and inactive kinase structures. The threading function was used to estimate the free energy cost to reorganize kinases from the active to classical inactive conformation, and we showed that this estimator is consistent with the results of molecular dynamics free energy simulations for a small set of STKs and TKs. In the current study, we analyze the results of a large-scale study of the binding affinities of 50 type II inhibitors to 348 kinases, of which the results for 16 of the 50 type II inhibitors were reported in an earlier study (the "Davis dataset"); the binding data for the remaining 34 type II inhibitors to the panel of 348 kinases were recently obtained (the "Schrodinger dataset"). We use the Potts statistical energy model to investigate the contribution of protein reorganization to the selectivity of the large kinase panel against the set of 50 type II inhibitors, and find that protein reorganization makes a significant contribution to the selectivity. The AUC of the receiver-operator characteristic curve is [~]0.8. We report the results of an internal "blind test", that shows how Potts threading energies can provide more accurate estimates of kinase selectivity than corresponding predictions using experimental results of small sample size. We discuss why two STK phylogenetic kinase families, STE and CMGC, appear to contain many outliers, and how to improve the ability to predict kinase selectivity with a more complete analysis of the kinase conformational landscape. We compare the performance of Potts threading for predicting binding properties of the large set of (50) Type II inhibitors to 348 kinases, with those of a sequence-based purely machine learning model, DeepDTAGen, a publicly available machine learning model that was trained on the complete Davis dataset, including both Type I and Type II kinase inhibitors. We observe that DeepDTAGen performs well on binding predictions for the 16 type II inhibitors in the Davis dataset, but performs poorly on binding predictions for the 34 type II inhibitors against 348 kinases in the Schrodinger dataset.

We introduce AI-BioMech, a deep learning based framework that directly predicts the mechanical response of cellular structures from 2D images, eliminating the need for manual geometry definition and traditional finite element simulations. The framework is trained on synthetic datasets representing biological cellular structures and benchmarked against real experimental data. Finite element analysis (FEA) based labeling is used to generate pixel level annotations for semantic segmentation, enabling accurate identification of stress and strain distributions. By learning spatial and hierarchical patterns from these annotations, the model automatically extracts complex features to predict cellular material responses under compressive loading conditions. Transfer learning with fine tuning by using the DeepLabv3 architecture with ResNet50, ResNet101, and Inception ResNetV2 backbones enhances prediction accuracy and generalization from limited datasets. Model predictions are validated against experimental results and Digital Image Correlation (DIC) measurements, demonstrating strong agreement with physical observations. The results show that AI-BioMech achieves up to 99% prediction accuracy while significantly outperforming traditional methods in computational speed and scalability.