Molecular Systems Design & Engineering最新文献

Accurate predictive thermodynamic models are essential tools for the computational design of new molecules. Models based on COnductor-like Screening Models (COSMO), such as COSMO-SAC and COSMO-RS, are well-suited for this purpose but require computationally expensive quantum-mechanical (QM) calculations to generate key properties - the surface charge density distribution (p(σ)), the surface area (A) and the cavity volume (V) - for each molecule. Here we develop graph-based neural network surrogate models, based on a directed message passing neural network (DMPNN) and a graph convolutional network (GCN), to predict these molecular properties rapidly and accurately. We train the models on a dataset of over 16 000 compounds generated with an automated QM calculation pipeline. The DMPNN model outperforms GCN for the prediction of the surface charge density, while the GCN provides a higher accuracy for surface area and cavity volume. We build on these strengths to propose a hybrid model: COSMO-NET. When applied to predict octanol-water partition coefficients and solid-liquid solubility across diverse solvents, COSMO-NET consistently outperforms individual models with the ranking: COSMO-NET > DMPNN > GCN > ECFP-MLP. COSMO-NET demonstrates remarkable robustness with minimal performance variation across different data splitting strategies. These results demonstrate that machine-learning surrogates can reliably replace costly QM calculations while maintaining accuracy, enabling the rapid discovery of new molecules and accurate evaluation of their thermodynamic properties.

Adsorption-based separation using metal-organic frameworks (MOFs) represents a promising, energy-efficient alternative to conventional separation technologies. While the vast design space of MOFs allows for precise property tuning to specific applications, the sheer scale of the design space renders exhaustive experimental testing impractical. Consequently, materials discovery relies on large-scale computational screening. Current screening workflows typically employ Grand-Canonical Monte Carlo (GCMC) simulations to predict adsorption properties; however, the high computational cost of GCMC often limits the number of materials screened. In this work, we introduce a process-level screening paradigm leveraging the computational speed of GPU-accelerated 3D classical density functional theory (cDFT). Validation against state-of-the-art GCMC simulations demonstrates that cDFT accurately predicts process key performance indicators (KPIs). The KPIs calculated via cDFT generally deviate by less than 5% from GCMC results while reducing computational costs by two to four orders of magnitude. Leveraging this increased efficiency, we screen the publicly available CoRE MOF 2025 database for methane/nitrogen separation using a temperature-swing adsorption process. By evaluating performance across a wide range of feed compositions, from industrial gas streams to dilute methane sources, the study identifies MOF candidates with robust performance profiles. The entire screening of over 5000 MOFs, requiring 460 000 adsorption calculations, was completed in only five days using two GPUs. The work establishes cDFT as a reliable, high-throughput approach for process-informed material discovery.

Chimeric antigen receptor (CAR) T-cell migration and infiltration into solid tumours remains a central challenge for cancer immunotherapy. While factors such as chemokine gradients, extracellular matrix (ECM) stiffness, and immunosuppressive signalling have been well-documented, the physical confinement imposed by tumour architecture is increasingly recognized as a critical barrier. Tumour-associated ECM remodelling generates micro- and nano-scale geometric constraints that significantly alter how immune cells, particularly CAR T-cells, navigate the tumour microenvironment (TME). This article argues that physical confinement merits to be positioned at the centre of the migration challenge, not as an accessory to chemical or biochemical cues. Sustained from both fundamental cell migration biology and recent engineering advances, we highlight how designing microfluidic devices with defined geometric features can uniquely capture the role of confinement in immune cell behaviour and model this challenge. By enabling precise design control over pore size, channel geometry, and nature-inspired structures, these platforms will allow researchers to decouple confinement from other overlapping variables such as stiffness.

Advanced functions in biological tissues depend strongly on the hierarchical architecture and orientation of polymer networks rather than on composition alone. Although anisotropic hydrogels have been engineered using stretching, external-field alignment, and 3D printing, many approaches require external energy input or complex processing. Here, inspired by the biomineralization principle of template-driven local enrichment and structural transfer, we extend this concept to an organic polymer–organic polymer system and demonstrate hierarchical inheritance of anisotropic architectures in interpenetrating polymer network (IPN) hydrogels via a multi-stage soft-templating strategy driven solely by intermolecular interactions. An oriented gelatin network was first self-organized on a polypropylene (PP) substrate and subsequently combined with poly(acrylic acid) (PAAc) to form gelatin/PAAc IPN hydrogels through UV polymerization of acrylic acid (AAc). During dialysis, the IPN hydrogels exhibited characteristic turbidity followed by shrinkage and gradual clarification from the periphery toward the center. TG–DTA analysis revealed significantly reduced water content and increased polymer fraction compared with single-network hydrogels. Although clear anisotropic features were not fully resolved at the microscopic level, multiple macroscopic and structural analyses consistently indicated anisotropic network organization. PP-templated IPNs exhibited consistently higher apparent compressive moduli than glass-templated IPNs and showed anisotropic thermoresponsive deformation with distinct swelling and contraction along the diameter and thickness directions. Control IPNs incorporating PNIPAm as the second network did not exhibit pronounced anisotropy or comparable strengthening, indicating that selective intermolecular interactions are essential for anisotropy inheritance. This study introduces structural information inheritance between organic polymer networks and provides a general framework for energy-free hierarchical programming of soft materials with directionally tunable mechanical and stimuli-responsive properties.

A D–A copolymer, poly(4,8-bis(5-(2-ethylhexyl)thiophen-2-yl)benzo[1,2-b:4,5-b']dithiophene-2,6-diyl)-co-(9-(2-octyldodecyl)-8H-pyrrolo[3,4-b]dithieno[2,3-f:3′,2′-h]quinoxaline-8,10(9H)-dione), denoted as P(BDTPtQD)—consisting of dithienoquinoxalineimide (DTQxI) as the acceptor (A) and benzodithiophene with thiophene-conjugated side chain (BDTT) as the donor (D) unit—was synthesized and its optical and electrochemical properties were compared with those of the well-known polymer PBDB-T, which exhibits the same donor unit and a different acceptor unit. P(BDTPtQD) exhibits a large dipole moment, a wide bandgap of 1.87 eV, and a deeper highest occupied molecular orbital (HOMO) energy level of −5.45 eV as compared to PBDB-T. When combined with narrow bandgap non-fullerene acceptor BTP-eC9 and medium bandgap non-fullerene acceptor DBTBT-IC, the optimized binary organic solar cells showed an overall power conversion efficiencies of 14.84 and 12.81%, respectively. Following the addition of DBTBT-IC to the P(BDTPtQD) : BTP-eC9 binary host, ternary organic solar cells fabricated under air-processed conditions achieved a power conversion efficiency of 17.26%. The improvement in the power conversion efficiency of the ternary organic solar cell is attributed to the denser π–π stacking distance, the longer crystal coherence length of π–π stacking, efficient exciton utilization through energy transfer from DBTBT-IC to BTP-eC9, lower trap density, and minimized recombination losses.

The vinylpyrrolidone–vinyl acetate copolymer (copovidone) is commonly used in oral solid dosage systems for its solubilising and binding properties, but suffers from native hygroscopicity and the high mechanical fibre strength of its extruded strands, resulting in processing issues, compromised manufacturability, and formulation issues during long-term storage. Herein, we report a co-processed excipient, copovidone (CPS) (Plasdone™ S-630) with colloidal silicon dioxide (Aerosil® 200 Pharma) (AP), processed using a twin-screw HME setup. AP was used to counteract moisture sensitivity and minimise extrude hardness by creating brittleness without changing the chemical composition. The optimised extrudes were extensively characterised by texture analysis, differential scanning calorimetry (DSC), thermogravimetric analysis (TGA), hot stage microscopy (HSM), contact angle and wettability studies, Fourier-transform infrared spectroscopy (FTIR), X-ray diffraction (XRD), scanning electron microscopy (SEM), and X-ray micro-computed tomography. Ritonavir (RTV), a poorly water-soluble (BCS class II) antiretroviral agent, was used as a model drug to assess the functional performance of the fabricated excipients in direct-compression tablet formulations. Tablets with a co-processed excipient exhibited a 1.70-fold increase in % drug release at 2 h (97% in 120 minutes). This work presents a facile method for designing a suitable co-processed excipient capable of modulating the release of poorly aqueous-soluble drugs.

Molecularly imprinted polymers (MIPs) are synthetic polymer networks formed around template molecules to create selective molecular recognition sites. While decades of research have advanced our understanding of pre-polymerisation complexation and the properties of finished MIP materials, the polymerisation step itself remains a conceptual black box in imprinting design. This Perspective argues that polymerisation—the dynamic process by which monomers are covalently linked and cross-linked around a template—is the decisive yet under-explored stage governing the fidelity, selectivity, and function of imprinted polymer networks. We adopt a design-first, mechanism-centric viewpoint to examine how polymerisation encodes molecular recognition and catalytic function in MIPs, molecularly imprinted catalysts (MICs), and enantiopure molecularly imprinted catalysts (EMICs). We explain why conventional fixed-topology molecular simulations, such as classical molecular dynamics, are intrinsically unable to capture imprinting fidelity, and we highlight how reactive molecular dynamics approaches open new avenues for resolving cavity formation, network growth, and the emergence of functional sites. Polymerisation is reframed as a dynamic encoding process in which the trajectory of network formation—the sequence of radical initiation, propagation, and cross-linking events—critically determines the architecture and performance of binding sites. We propose mechanistic descriptors of polymer growth, including growth-front propagation patterns, cross-linking motifs, and kinetic trapping phenomena, as new design variables for imprinting that move beyond static pre-assembly metrics. Insights from MIPs are generalised to MICs and EMICs, in which polymerisation locks in transition-state-stabilising functional-group arrangements, and implications for other porous polymer and molecular-recognition materials are briefly discussed. By elucidating polymerisation mechanisms, this Perspective establishes a broadly applicable, predictive, mechanism-driven framework for the rational design of imprinted polymer networks and calls for polymerisation to be treated not as an uncontrollable curing step but as a tunable, information-rich process central to the next generation of molecularly imprinted materials.

A graphical abstract is available for this content

Rapid and accurate prediction of the colloidal phase diagrams is necessary for controlling their structure and properties. Mapping these multi-dimensional phase diagrams with molecular dynamics (MD) is costly, especially when sharp, localized transition boundaries demand dense sampling to resolve the boundaries. We target this challenge by actively learning the boundary where the order parameter – the number of nearest neighbours, N_nn – undergoes an abrupt change. We describe colloids interacting via isotropic pair potentials with tunable range of interaction and aim to identify the boundary between the dilute and condensed phases. We model N_nn = f(x) with a Gaussian process (GP) and derive a novel gradient-aware acquisition strategy that prioritizes locations with large and/or uncertain spatial derivatives of the GP posterior. The gradient-aware strategy is based on the thermodynamic principle of large susceptibility, i.e., large derivative of the order parameter, at phase boundaries. A simple ellipse-based exclusion heuristic helps spread acquisitions along the boundary manifold. Across 1D and 2D planes of the phase diagram, the proposed strategy localizes phase boundaries accurately with significantly fewer MD runs than dense grid search and outperforms random acquisition or standard acquisition strategies unaware of the gradient of the GP. The framework can be readily extended to higher-dimensional space and support multi-fidelity MD simulations, providing a practical route to sample-efficient and physics-aligned discovery of colloidal phase diagrams.

Self-assembled nanostructures comprising Phenylalanine-based oligopeptides are widely utilized in various applications across the domains of electronics, personal care, and biomedicine. Numerous factors, including sequence, determine the morphology of these nanostructures. Computational approaches such as bottom-up coarse-grained (CG) models coupled with the molecular dynamics simulation technique can be used to predict morphologies as a function of the sequence. These approaches efficiently resolve the multiple scales relevant to the self-assembly process while generating CG representations that are consistent with the all-atom representations. However, bottom-up CG models are computationally demanding to develop. This study proposes an approach to develop bottom-up CG models of Phenylalanine oligopeptides that are transferable between sequences to resolve their assembly. The approach builds upon a previously reported bottom-up CG model for the Phenylalanine tripeptide to develop parameters for new degrees of freedom associated with other oligopeptides. The performance of the new CG model is analyzed by comparing simulation results with corresponding results obtained using different versions of the Martini force field. The predictions from the new force field are also compared with available experimental results.