Predicting protein-ligand binding affinity is a fundamental challenge in drug discovery. Recent advances in deep learning have led to the development of numerous models, many of which rely on three-dimensional protein-ligand complex structures and focus primarily on affinity prediction. In this study, we introduce an Algebraic Graph Neural Network (AGNN) model designed to encode molecular structures into a low-dimensional graph representation while preserving critical biochemical interactions. While algebraic graph theory has been widely used in physical modeling and molecular studies, traditional methods often struggle to accurately capture the complexity of biomolecular interactions. To address this limitation, our proposed AGNN model leverages multiscale weighted colored subgraphs to describe molecular interactions through a graph neural network. These representations allow the model to effectively learn the geometric and topological features of protein-ligand complexes. The AGNN model integrates graph convolution layers and attention mechanisms to refine feature extraction and improve the interpretability of learned embeddings. Furthermore, we incorporate gradient boosting decision trees (GBDTs) to enhance the prediction of binding affinities by capturing nonlinear relationships between molecular features. Our approach is extensively validated using benchmark datasets, including PDBBind and CASF-2016, demonstrating superior performance in binding affinity prediction compared to state-of-the-art scoring functions.
This study focusses on one-dimensional (1D) modeling of transient transdermal drug delivery across the epidermis via stratum corneum. In order to mimic in this manuscript how drugs pass through the skin, Fick’s second law has been incorporated as a flow governing equation in microneedles and is solved using a Lattice Boltzmann model. In this study we have assumed that the contact region between skin and drug as a point. To depict this phenomenon, the effect of parameters such as time duration, diffusion coefficient, length of the microneedle, initial drug concentration, cumulative flux and fraction of the drug delivered has been investigated. To validate the present model, the obtained results are compared with an existing analytical results.
Blood is a multifaceted colloidal suspension of plasma as an aqueous electrolyte-protein fluid and cellular components, mainly red blood cells, the rheology of which is influenced by mutually reinforcing hydrodynamic and chemical interactions, such as protein-mediated aggregation, plasma composition, and solute transport. These physicochemical interactions are magnified in stenosed micro vessels resulting in great changes in effective viscosity and haematocrit distribution. The present study develops a theoretical two-layer model in order to explain the synergistic action of stenosis and microcirculation on Jeffrey fluid dynamics. The middle region is modelled as a viscoelastic Jeffrey fluid, which considers the red blood cell elasticity and shear-thinning effect, and the outer layer as a Newtonian plasma. The governing equations are analytically solved with slip boundary conditions and mild stenosis with analytic expressions of velocity profiles, flow flux, effective viscosity, core haematocrit, and mean haematocrit. The findings indicate that an effective viscosity is positively proportional to stenosis height, channel half-width and slip parameter and negatively related to the Jeffrey parameter, and thus, the interplay between the viscous recovery of red blood cells and the breakup of plasma-protein networks induced by shear is demonstrated. Moreover, core and mean haematocrit drop off with a growing degree of stenosis and slip, as predicted by the Fahraeus-Lindqvist effect and the law of colloidal segregation. The suggested framework combines rheological modelling and chemical understanding connecting the flow changes caused by stenosis with the transport of solutes, reaction-diffusion dynamics, and drug and nanoparticle delivery efficacy, which provides a chemically informed view of hemorheology with potential biomedical chemistry, pharmaceutical formulation, and clinical diagnostics applications.
Three-phase fluid-structure interaction (FSI) problems couple the motion of solid structures and two different fluids, often exhibiting complex dynamics and posing challenges for computational investigation. The main objective of this paper is to present a new computational framework for the modeling, simulation, and application of such problems. To that end, we propose an integrated numerical algorithm that utilizes the volume of fluid method for the two-fluid flow and the immersed boundary method for the fluid-solid interaction, with a special emphasis on three-phase FSI computation in micro-scale settings. To demonstrate our methodology, we conduct a pilot FSI study on cellulose microfibrils in plant cell walls by simulating the interaction between microfibrils and two fluids, pectin and water. We investigate the evolution of the flow fields, the speed and spatial organization of the microfibrils, and the impact of different initial settings on microfibril dynamics. We find that the microfibril-pectin-water interaction leads to transversely oriented microfibrils with respect to the elongation axis of the plant cell. The simulation results indicate that our methodology may provide a novel approach to investigate the complex behavior of microfibrils and gain insights into the intrinsic dynamical properties of such microstructures.
The research on cancer cells has proposed many hypotheses as a basic conceptual framework for understanding how normal cells transform into cancer cells. Cancer cells are assumed to undergo a multilevel mutation process from normal cells to cancer cells. A kinetics model of the multistep transformation of normal cells into cancer cells was developed to provide insight into the fundamental aspects of cancer cell evolution. The kinetic model is composed of coupled ordinary differential equations to describe the mutation process. This combination of equations can describe how mutational processes, such as angiogenesis, cell death rates, genetic instability, and replication rates. The main predictions of the multistep mutation model are based on the fastest-growing cancer cells. The new estimation method was used in this study to estimate the parameters in the kinetics model. Parameters were estimated using the particle swarm optimization algorithm. A new technique in this study, parameters are estimated to obtain the possible contribution of each parameter to the fastest-growing cancer cells. The results of this study present a new estimate of the multilevel mutation process based on parameters from breast cancer data.
