A generalized deep learning approach for local structure identification in molecular simulations

Machine Learning Classification of Local Environments in Molecular Crystals

Identifying local structural motifs and packing patterns of molecular solids is a challenging task for both simulation and experiment. We demonstrate two novel approaches to characterize local environments in different polymorphs of molecular crystals using learning models that employ either flexibly learned or handcrafted molecular representations. In the first case, we follow our earlier work on graph learning in molecular crystals, deploying an atomistic graph convolutional network combined with molecule-wise aggregation to enable per-molecule environmental classification. For the second model, we develop a new set of descriptors based on symmetry functions combined with a point-vector representation of the molecules, encoding information about the positions and relative orientations of the molecule. We demonstrate very high classification accuracy for both approaches on urea and nicotinamide crystal polymorphs and practical applications to the analysis of dynamical trajectory data for nanocrystals and solid-solid interfaces. Both architectures are applicable to a wide range of molecules and diverse topologies, providing an essential step in the exploration of complex condensed matter phenomena.

In-layer inhomogeneity of molecular dynamics in quasi-liquid layers of ice

Quasi-liquid layers (QLLs) are present on the surface of ice and play a significant role in its distinctive chemical and physical properties. These layers exhibit considerable heterogeneity across different scales ranging from nanometers to millimeters. Although the formation of partially ice-like structures has been proposed, the molecular-level understanding of this heterogeneity remains unclear. Here, we examined the heterogeneity of molecular dynamics on QLLs based on molecular dynamics simulations and machine learning analysis of the simulation data. We demonstrated that the molecular dynamics of QLLs do not comprise a mixture of solid- and liquid water molecules. Rather, molecules having similar behaviors form dynamical domains that are associated with the dynamical heterogeneity of supercooled water. Nonetheless, molecules in the domains frequently switch their dynamical state. Furthermore, while there is no observable characteristic domain size, the long-range ordering strongly depends on the temperature and crystal face. Instead of a mixture of static solid- and liquid-like regions, our results indicate the presence of heterogeneous molecular dynamics in QLLs, which offers molecular-level insights into the surface properties of ice.

Classification of complex local environments in systems of particle shapes through shape symmetry-encoded data augmentation

Detecting and analyzing the local environment is crucial for investigating the dynamical processes of crystal nucleation and shape colloidal particle self-assembly. Recent developments in machine learning provide a promising avenue for better order parameters in complex systems that are challenging to study using traditional approaches. However, the application of machine learning to self-assembly on systems of particle shapes is still underexplored. To address this gap, we propose a simple, physics-agnostic, yet powerful approach that involves training a multilayer perceptron (MLP) as a local environment classifier for systems of particle shapes, using input features such as particle distances and orientations. Our MLP classifier is trained in a supervised manner with a shape symmetry-encoded data augmentation technique without the need for any conventional roto-translations invariant symmetry functions. We evaluate the performance of our classifiers on four different scenarios involving self-assembly of cubic structures, two-dimensional and three-dimensional patchy particle shape systems, hexagonal bipyramids with varying aspect ratios, and truncated shapes with different degrees of truncation. The proposed training process and data augmentation technique are both straightforward and flexible, enabling easy application of the classifier to other processes involving particle orientations. Our work thus presents a valuable tool for investigating self-assembly processes on systems of particle shapes, with potential applications in structure identification of any particle-based or molecular system where orientations can be defined.

Enhanced Sampling of Crystal Nucleation with Graph Representation Learnt Variables

In this study, we present a graph neural network (GNN)-based learning approach using an autoencoder setup to derive low-dimensional variables from features observed in experimental crystal structures. These variables are then biased in enhanced sampling to observe state-to-state transitions and reliable thermodynamic weights. In our approach, we used simple convolution and pooling methods. To verify the effectiveness of our protocol, we examined the nucleation of various allotropes and polymorphs of iron and glycine in their molten states. Our graph latent variables, when biased in well-tempered metadynamics, consistently show transitions between states and achieve accurate thermodynamic rankings in agreement with experiments, both of which are indicators of dependable sampling. This underscores the strength and promise of our GNN variables for improved sampling. The protocol shown here should be applicable for other systems and other sampling methods.

Graph-Neural-Network-Based Unsupervised Learning of the Temporal Similarity of Structural Features Observed in Molecular Dynamics Simulations

Classification of molecular structures is a crucial step in molecular dynamics (MD) simulations to detect various structures and phases within systems. Molecular structures, which are commonly identified using order parameters, were recently identified using machine learning (ML), that is, the ML models acquire structural features using labeled crystals or phases via supervised learning. However, these approaches may not identify unlabeled or unknown structures, such as the imperfect crystal structures observed in nonequilibrium systems and interfaces. In this study, we proposed the use of a novel unsupervised learning framework, denoted temporal self-supervised learning (TSSL), to learn structural features and design their parameters. In TSSL, the ML models learn that the structural similarity is learned via contrastive learning based on minor short-term variations caused by perturbations in MD simulations. This learning framework is applied to a sophisticated architecture of graph neural network models that use bond angle and length data of the neighboring atoms. TSSL successfully classifies water and ice crystals based on high local ordering, and furthermore, it detects imperfect structures typical of interfaces such as the water-ice and ice-vapor interfaces.

Advances in Computational Approaches for Estimating Passive Permeability in Drug Discovery

Passive permeation of cellular membranes is a key feature of many therapeutics. The relevance of passive permeability spans all biological systems as they all employ biomembranes for compartmentalization. A variety of computational techniques are currently utilized and under active development to facilitate the characterization of passive permeability. These methods include lipophilicity relations, molecular dynamics simulations, and machine learning, which vary in accuracy, complexity, and computational cost. This review briefly introduces the underlying theories, such as the prominent inhomogeneous solubility diffusion model, and covers a number of recent applications. Various machine-learning applications, which have demonstrated good potential for high-volume, data-driven permeability predictions, are also discussed. Due to the confluence of novel computational methods and next-generation exascale computers, we anticipate an exciting future for computationally driven permeability predictions.

Controlling crystallization: what liquid structure and dynamics reveal about crystal nucleation mechanisms

Over recent years, molecular simulations have provided invaluable insights into the microscopic processes governing the initial stages of crystal nucleation and growth. A key aspect that has been observed in many different systems is the formation of precursors in the supercooled liquid that precedes the emergence of crystalline nuclei. The structural and dynamical properties of these precursors determine to a large extent the nucleation probability as well as the formation of specific polymorphs. This novel microscopic view on nucleation mechanisms has further implications for our understanding of the nucleating ability and polymorph selectivity of nucleating agents, as these appear to be strongly linked to their ability in modifying structural and dynamical characteristics of the supercooled liquid, namely liquid heterogeneity. In this perspective, we highlight recent progress in exploring the connection between liquid heterogeneity and crystallization, including the effects of templates, and the potential impact for controlling crystallization processes. This article is part of a discussion meeting issue 'Supercomputing simulations of advanced materials'.

Rapid Prediction of a Liquid Structure from a Single Molecular Configuration Using Deep Learning

Molecular dynamics simulation is an indispensable tool for understanding the collective behavior of atoms and molecules and the phases they form. Statistical mechanics provides accurate routes for predicting macroscopic properties as time-averages over visited molecular configurations - microstates. However, to obtain convergence, we need a sufficiently long record of visited microstates, which translates to the high-computational cost of the molecular simulations. In this work, we show how to use a point cloud-based deep learning strategy to rapidly predict the structural properties of liquids from a single molecular configuration. We tested our approach using three homogeneous liquids with progressively more complex entities and interactions: Ar, NO, and H₂O under varying pressure and temperature conditions within the liquid state domain. Our deep neural network architecture allows rapid insight into the liquid structure, here probed by the radial distribution function, and can be used with molecular/atomistic configurations generated by either simulation, first-principle, or experimental methods.

A Universal and Accurate Method for Easily Identifying Components in Raman Spectroscopy Based on Deep Learning

Raman spectroscopy has been widely used to provide the structural fingerprint for molecular identification. Due to interference from coexisting components, noise, baseline, and systematic differences between spectrometers, component identification with Raman spectra is challenging, especially for mixtures. In this study, a method entitled DeepRaman has been proposed to solve those problems by combining the comparison ability of a pseudo-Siamese neural network (pSNN) and the input-shape flexibility of spatial pyramid pooling (SPP). DeepRaman was trained, validated, and tested with 41,564 augmented Raman spectra from two databases (pharmaceutical material and S.T. Japan). It can achieve 96.29% accuracy, 98.40% true positive rate (TPR), and 94.36% true negative rate (TNR) on the test set. Another six data sets measured on different instruments were used to evaluate the performance of the proposed method from different aspects. DeepRaman can provide accurate identification results and significantly outperform the hit quality index (HQI) method and other deep learning models. In addition, it performs well in cases of different spectral complexity and low-content components. Once the model is established, it can be used directly on different data sets without retraining or transfer learning. Furthermore, it also obtains promising results for the analysis of surface-enhanced Raman spectroscopy (SERS) data sets and Raman imaging data sets. In summary, it is an accurate, universal, and ready-to-use method for component identification in various application scenarios.

Phase diagrams-Why they matter and how to predict them

Understanding the thermodynamic stability and metastability of materials can help us to, for example, gauge whether crystalline polymorphs in pharmaceutical formulations are likely to be durable. It can also help us to design experimental routes to novel phases with potentially interesting properties. In this Perspective, we provide an overview of how thermodynamic phase behavior can be quantified both in computer simulations and machine-learning approaches to determine phase diagrams, as well as combinations of the two. We review the basic workflow of free-energy computations for condensed phases, including some practical implementation advice, ranging from the Frenkel-Ladd approach to thermodynamic integration and to direct-coexistence simulations. We illustrate the applications of such methods on a range of systems from materials chemistry to biological phase separation. Finally, we outline some challenges, questions, and practical applications of phase-diagram determination which we believe are likely to be possible to address in the near future using such state-of-the-art free-energy calculations, which may provide fundamental insight into separation processes using multicomponent solvents.

Collective Variables for Crystallization Simulations-from Early Developments to Recent Advances

Crystallization is an important physicochemical process which has relevance in material science, biology, and the environment. Decades of experimental and theoretical efforts have been made to understand this fundamental symmetry-breaking transition. While experiments provide equilibrium structures and shapes of crystals, they are limited to unraveling how molecules aggregate to form crystal nuclei that subsequently transform into bulk crystals. Computer simulations, mainly molecular dynamics (MD), can provide such microscopic details during the early stage of a crystallization event. Crystallization is a rare event that takes place in time scales much longer than a typical equilibrium MD simulation can sample. This inadequate sampling of the MD method can be easily circumvented by the use of enhanced sampling (ES) simulations. In most of the ES methods, the fluctuations of a system's slow degrees of freedom, called collective variables (CVs), are enhanced by applying a bias potential. This transforms the system from one state to the other within a short time scale. The most crucial part of such CV-based ES methods is to find suitable CVs, which often needs intuition and several trial-and-error optimization steps. Over the years, a plethora of CVs has been developed and applied in the study of crystallization. In this review, we provide a brief overview of CVs that have been developed and used in ES simulations to study crystallization from melt or solution. These CVs can be categorized mainly into four types: (i) spherical particle-based, (ii) molecular template-based, (iii) physical property-based, and (iv) CVs obtained from dimensionality reduction techniques. We present the context-based evolution of CVs, discuss the current challenges, and propose future directions to further develop effective CVs for the study of crystallization of complex systems.

A method to create real-like point clouds for 3D object classification

There are a large number of publicly available datasets of 3D data, they generally suffer from some drawbacks, such as small number of data samples, and class imbalance. Data augmentation is a set of techniques that aim to increase the size of datasets and solve such defects, and hence to overcome the problem of overfitting when training a classifier. In this paper, we propose a method to create new synthesized data by converting complete meshes into occluded 3D point clouds similar to those in real-world datasets. The proposed method involves two main steps, the first one is hidden surface removal (HSR), where the occluded parts of objects surfaces from the viewpoint of a camera are deleted. A low-complexity method has been proposed to implement HSR based on occupancy grids. The second step is a random sampling of the detected visible surfaces. The proposed two-step method is applied to a subset of ModelNet40 dataset to create a new dataset, which is then used to train and test three different deep-learning classifiers (VoxNet, PointNet, and 3DmFV). We studied classifiers performance as a function of the camera elevation angle. We also conducted another experiment to show how the newly generated data samples can improve the classification performance when they are combined with the original data during training process. Simulation results show that the proposed method enables us to create a large number of new data samples with a small size needed for storage. Results also show that the performance of classifiers is highly dependent on the elevation angle of the camera. In addition, there may exist some angles where performance degrades significantly. Furthermore, data augmentation using our created data improves the performance of classifiers not only when they are tested on the original data, but also on real data.

A Deep Learning Framework Discovers Compositional Order and Self-Assembly Pathways in Binary Colloidal Mixtures

Binary colloidal superlattices (BSLs) have demonstrated enormous potential for the design of advanced multifunctional materials that can be synthesized via colloidal self-assembly. However, mechanistic understanding of the three-dimensional self-assembly of BSLs is largely limited due to a lack of tractable strategies for characterizing the many two-component structures that can appear during the self-assembly process. To address this gap, we present a framework for colloidal crystal structure characterization that uses branched graphlet decomposition with deep learning to systematically and quantitatively describe the self-assembly of BSLs at the single-particle level. Branched graphlet decomposition is used to evaluate local structure via high-dimensional neighborhood graphs that quantify both structural order (e.g., body-centered-cubic vs face-centered-cubic) and compositional order (e.g., substitutional defects) of each individual particle. Deep autoencoders are then used to efficiently translate these neighborhood graphs into low-dimensional manifolds from which relationships among neighborhood graphs can be more easily inferred. We demonstrate the framework on in silico systems of DNA-functionalized particles, in which two well-recognized design parameters, particle size ratio and interparticle potential well depth can be adjusted independently. The framework reveals that binary colloidal mixtures with small interparticle size disparities (i.e., A- and B-type particle radius ratios of r _A/r _B = 0.8 to r _A/r _B = 0.95) can promote the self-assembly of defect-free BSLs much more effectively than systems of identically sized particles, as nearly defect-free BCC-CsCl, FCC-CuAu, and IrV crystals are observed in the former case. The framework additionally reveals that size-disparate colloidal mixtures can undergo nonclassical nucleation pathways where BSLs evolve from dense amorphous precursors, instead of directly nucleating from dilute solution. These findings illustrate that the presented characterization framework can assist in enhancing mechanistic understanding of the self-assembly of binary colloidal mixtures, which in turn can pave the way for engineering the growth of defect-free BSLs.

Stabilizing a Double Gyroid Network Phase with 2 nm Feature Size by Blending of Lamellar and Cylindrical Forming Block Oligomers

Molecular dynamics simulations are used to study binary blends of an AB-type diblock and an AB₂-type miktoarm triblock amphiphiles (also known as high-χ block oligomers) consisting of sugar-based (A) and hydrocarbon (B) blocks. In their pure form, the AB diblock and AB₂ triblock amphiphiles self-assemble into ordered lamellar (LAM) and cylindrical (CYL) structures, respectively. At intermediate compositions, however, the AB₂-rich blend (0.2 ≤ x _AB ≤ 0.4) forms a double gyroid (DG) network, whereas perforated lamellae (PL) are observed in the AB-rich blend (0.5 ≤ x _AB ≤ 0.8). All of the ordered mesophases present domain pitches under 3 nm, with 1 nm feature sizes for the polar domains. Structural analyses reveal that the nonuniform interfacial curvatures of DG and PL structures are supported by local composition variations of the LAM- and CYL-forming amphiphiles. Self-consistent mean field theory calculations for blends of related AB and AB₂ block polymers also show the DG network at intermediate compositions, when A is the minority block, but PL is not stable. This work provides molecular-level insights into how blending of shape-filling molecular architectures enables network phase formation with extremely small feature sizes over a wide composition range.

Investigating the quasi-liquid layer on ice surfaces: a comparison of order parameters

Ice surfaces are characterized by pre-melted quasi-liquid layers (QLLs), which mediate both crystal growth processes and interactions with external agents. Understanding QLLs at the molecular level is necessary to unravel the mechanisms of ice crystal formation. Computational studies of the QLLs heavily rely on the accuracy of the methods employed for identifying the local molecular environment and arrangements, discriminating between solid-like and liquid-like water molecules. Here we compare the results obtained using different order parameters to characterize the QLLs on hexagonal ice (Ih) and cubic ice (Ic) model surfaces investigated with molecular dynamics (MD) simulations in a range of temperatures. For the classification task, in addition to the traditional Steinhardt order parameters in different flavours, we select an entropy fingerprint and a deep learning neural network approach (DeepIce), which are conceptually different methodologies. We find that all the analysis methods give qualitatively similar trends for the behaviours of the QLLs on ice surfaces with temperature, with some subtle differences in the classification sensitivity limited to the solid-liquid interface. The thickness of QLLs on the ice surface increases gradually as the temperature increases. The trends of the QLL size and of the values of the order parameters as a function of temperature for the different facets may be linked to surface growth rates which, in turn, affect crystal morphologies at lower vapour pressure. The choice of the order parameter can be therefore informed by computational convenience except in cases where a very accurate determination of the liquid-solid interface is important.

Reverse vaccinology-based prediction of a multi-epitope SARS-CoV-2 vaccine and its tailoring to new coronavirus variants

The genome feature of SARS-CoV-2 leads the virus to mutate and creates new variants of concern. Tackling viral mutations is also an important challenge for the development of a new vaccine. Accordingly, in the present study, we undertook to identify B- and T-cell epitopes with immunogenic potential for eliciting responses to SARS-CoV-2, using computational approaches and its tailoring to coronavirus variants. A total of 47 novel epitopes were identified as immunogenic triggering immune responses and no toxic after investigation with in silico tools. Furthermore, we found these peptide vaccine candidates showed a significant binding affinity for MHC I and MHC II alleles in molecular docking investigations. We consider them to be promising targets for developing peptide-based vaccines against SARS-CoV-2. Subsequently, we designed two efficient multi-epitopes vaccines against the SARS-CoV-2, the first one based on potent MHC class I and class II T-cell epitopes of S (FPNITNLCPF-NYNYLYRLFR-MFVFLVLLPLVSSQC), M (MWLSYFIASF-GLMWLSYFIASFRLF), E (LTALRLCAY-LLFLAFVVFLLVTLA), and N (SPRWYFYYL-AQFAPSASAFFGMSR). The second candidate is the result of the tailoring of the first designed vaccine according to three classes of SARS-CoV-2 variants. Molecular docking showed that the protein-protein binding interactions between the vaccines construct and TLR2-TLR4 immune receptors are stable complexes. These findings confirmed that the final multi-epitope vaccine could be easily adapted to new viral variants. Our study offers a shortlist of promising epitopes that can accelerate the development of an effective and safe vaccine against the virus and its adaptation to new variants.Communicated by Ramaswamy H. Sarma.

Unsupervised topological learning for identification of atomic structures

We propose an unsupervised learning methodology with descriptors based on topological data analysis (TDA) concepts to describe the local structural properties of materials at the atomic scale. Based only on atomic positions and without a priori knowledge, our method allows for an autonomous identification of clusters of atomic structures through a Gaussian mixture model. We apply successfully this approach to the analysis of elemental Zr in the crystalline and liquid states as well as homogeneous nucleation events under deep undercooling conditions. This opens the way to deeper and autonomous study of complex phenomena in materials at the atomic scale.

Bond Orientational Order Parameters for Classifying Solid-like Clusters in a Lennard-Jones System near Liquid-Solid Transition and at Solid States

In this paper, we introduced an order parameter, named the local structure similarity (LSS), to measure the resemblance of a cluster structure in a liquid with respect to a perfect crystal. The LSS is based on a dot product of two bond orientational order complex vectors, with one vector associated with a particle in a liquid and the other vector with a particle in a crystal. The calculation of the LSS should scan the entire space of the Euler angles determined by the two coordinate frames describing individually the liquid and the crystal. The effectiveness of the LSS was examined by solid-like clusters in a Lennard-Jones (LJ) system near its liquid-solid phase transition and at solid states below its melting point, where the thermodynamic states of the LJ system were obtained by simulation annealing. The LSS measure was utilized to scrutinize the fcc-like, hcp-like, and bcc-like clusters classified by criteria based on W₄ and W₆ order parameters. As indicated by our results, the two ways of classification are consistent for fcc-like and hcp-like clusters, which are in a close resemblance to their crystalline counterparts. However, the classification with positive W₆ for bcc-like clusters is inconsistent with the results of the LSS measure, which was confirmed by clusters in a LJ system confined between two parallel slabs of particles in the bcc structure arrangement.

Descriptor-free unsupervised learning method for local structure identification in particle packings

Local structure identification is of great importance in many scientific and engineering fields. However, mathematical and supervised learning methods mostly rely on specific descriptors of local structure and can only be applied to particular packing configurations. In this work, we propose an improved unsupervised learning method, which is descriptor-free, for local structure identification in particle packing. The point cloud is used as the input of the improved method, which directly comes from spatial positions of particles and does not rely on specific descriptors. The improved method constructs an autoencoder based on the point cloud network combined with Gaussian mixture models for dimension reduction and clustering. Numerical examples show that the improved method performs well in local structure identification of quasicrystal disk and sphere packings, achieving comparable accuracy with previous methods. For disordered packings which have been considered nearly having no local structures, the improved method identifies a nontrivial 7-neighbor motif in the maximally dense random packing of disks, and finds acentric structural motifs in the random close packing of spheres, which demonstrate the ability on identification of new and unknown local structures. The improved unsupervised learning method would help mining information from massive simulation and experimental results, as well as devising new order parameters for particle packings.

BioS2Net: Holistic Structural and Sequential Analysis of Biomolecules Using a Deep Neural Network

BACKGROUND

For decades, the rate of solving new biomolecular structures has been exceeding that at which their manual classification and feature characterisation can be carried out efficiently. Therefore, a new comprehensive and holistic tool for their examination is needed.

METHODS

Here we propose the Biological Sequence and Structure Network (BioS2Net), which is a novel deep neural network architecture that extracts both sequential and structural information of biomolecules. Our architecture consists of four main parts: (i) a sequence convolutional extractor, (ii) a 3D structure extractor, (iii) a 3D structure-aware sequence temporal network, as well as (iv) a fusion and classification network.

RESULTS

We have evaluated our approach using two protein fold classification datasets. BioS2Net achieved a 95.4% mean class accuracy on the eDD dataset and a 76% mean class accuracy on the F184 dataset. The accuracy of BioS2Net obtained on the eDD dataset was comparable to results achieved by previously published methods, confirming that the algorithm described in this article is a top-class solution for protein fold recognition.

CONCLUSIONS

BioS2Net is a novel tool for the holistic examination of biomolecules of known structure and sequence. It is a reliable tool for protein analysis and their unified representation as feature vectors.

A point cloud-based deep learning strategy for protein-ligand binding affinity prediction

There is great interest to develop artificial intelligence-based protein-ligand binding affinity models due to their immense applications in drug discovery. In this paper, PointNet and PointTransformer, two pointwise multi-layer perceptrons have been applied for protein-ligand binding affinity prediction for the first time. Three-dimensional point clouds could be rapidly generated from PDBbind-2016 with 3772 and 11 327 individual point clouds derived from the refined or/and general sets, respectively. These point clouds (the refined or the extended set) were used to train PointNet or PointTransformer, resulting in protein-ligand binding affinity prediction models with Pearson correlation coefficients R = 0.795 or 0.833 from the extended data set, respectively, based on the CASF-2016 benchmark test. The analysis of parameters suggests that the two deep learning models were capable to learn many interactions between proteins and their ligands, and some key atoms for the interactions could be visualized. The protein-ligand interaction features learned by PointTransformer could be further adapted for the XGBoost-based machine learning algorithm, resulting in prediction models with an average Rp of 0.827, which is on par with state-of-the-art machine learning models. These results suggest that the point clouds derived from PDBbind data sets are useful to evaluate the performance of 3D point clouds-centered deep learning algorithms, which could learn atomic features of protein-ligand interactions from natural evolution or medicinal chemistry and thus have wide applications in chemistry and biology.

Mining of Effective Local Order Parameters to Classify Ice Polymorphs

Order parameters make it possible to quantify the degree of structural ordering in a material and thus to apply as the reaction coordinates during the free-energy analysis of phase or structure transitions. Furthermore, order parameters are useful in determining the local structures of molecular groups during transition stages. However, identifying or developing local order parameters (LOPs) that are sensitive for specific materials and phases is a non-trivial task. In this study, the ability of LOPs to classify the solid and liquid structures of water at coexistence or triple points is investigated with the aid of supervised machine learning. The classification accuracy of a total of 179,738,433 combinations of 493 LOPs is automatically and systematically compared for water structures at the ice Ih-Ic-liquid coexistence point and the ice III-V-liquid and ice V-VI-liquid triple points. The optimal sets of two LOPs are found for each point, and sets of three LOPs are suggested for better accuracy.

Searching for local order parameters to classify water structures at triple points

The diversity of ice polymorphs is of interest in condensed-matter physics, engineering, astronomy, and biosphere and climate studies. In particular, their triple points are critical to elucidate the formation of each phase and transitions among phases. However, an approach to distinguish their molecular structures is lacking. When precise molecular geometries are given, order parameters are often computed to quantify the degree of structural ordering and to classify the structures. Many order parameters have been developed for specific or multiple purposes, but their capabilities have not been exhaustively investigated for distinguishing ice polymorphs. Here, 493 order parameters and their combinations are considered for two triple points involving the ice polymorphs ice III-V-liquid and ice V-VI-liquid. Supervised machine learning helps automatic and systematic searching of the parameters. For each triple point, the best set of two order parameters was found that distinguishes three structures with high accuracy. A set of three order parameters is also suggested for better accuracy.

Semi-supervised learning for the study of structural formation in colloidal systems via image recognition

The analysis of the structural formation of colloidal systems using machine learning techniques has recently attracted much attention. In many of these studies, local bond-order parameters (LBOPs) were employed as descriptors, where such LBOPs are suitable mainly for the detection of crystal structures. On the other hand, image-based convolutional neural networks (CNNs) are quite effective in detecting not only crystals but also random structures, and the author demonstrated their efficiency in a previous paper. However, in supervised learning, it is difficult to obtain a correct result when there is an unexpected new phase that was unknown when training the CNN. In this paper, we propose a hybrid scheme that consists of supervised and unsupervised learning techniques, employing two different approaches: image-based CNN and generalized LBOP. The proposed method was applied to two-dimensional colloidal systems, and its efficiency was demonstrated.

From predictive modelling to machine learning and reverse engineering of colloidal self-assembly

An overwhelming diversity of colloidal building blocks with distinct sizes, materials and tunable interaction potentials are now available for colloidal self-assembly. The application space for materials composed of these building blocks is vast. To make progress in the rational design of new self-assembled materials, it is desirable to guide the experimental synthesis efforts by computational modelling. Here, we discuss computer simulation methods and strategies used for the design of soft materials created through bottom-up self-assembly of colloids and nanoparticles. We describe simulation techniques for investigating the self-assembly behaviour of colloidal suspensions, including crystal structure prediction methods, phase diagram calculations and enhanced sampling techniques, as well as their limitations. We also discuss the recent surge of interest in machine learning and reverse-engineering methods. Although their implementation in the colloidal realm is still in its infancy, we anticipate that these data-science tools offer new paradigms in understanding, predicting and (inverse) design of novel colloidal materials.

Development of a PointNet for Detecting Morphologies of Self-Assembled Block Oligomers in Atomistic Simulations

Molecular simulations with atomistic or coarse-grained force fields are a powerful approach for understanding and predicting the self-assembly phase behavior of complex molecules. Amphiphiles, block oligomers, and block polymers can form mesophases with different ordered morphologies describing the spatial distribution of the blocks, but entirely amorphous nature for local packing and chain conformation. Screening block oligomer chemistry and architecture through molecular simulations to find promising candidates for functional materials is aided by effective and straightforward morphology identification techniques. Capturing 3-dimensional periodic structures, such as ordered network morphologies, is hampered by the requirement that the number of molecules in the simulated system and the shape of the periodic simulation box need to be commensurate with those of the resulting network phase. Common strategies for structure identification include structure factors and order parameters, but these fail to identify imperfect structures in simulations with incorrect system sizes. Building upon pioneering work by DeFever et al. [Chem. Sci. 2019, 10, 7503-7515] who implemented a PointNet (i.e., a neural network designed for computer vision applications using point clouds) to detect local structure in simulations of single-bead particles and water molecules, we present a PointNet for detection of nonlocal ordered morphologies of complex block oligomers. Our PointNet was trained using atomic coordinates from molecular dynamics simulation trajectories and synthetic point clouds for ordered network morphologies that were absent from previous simulations. In contrast to prior work on simple molecules, we observe that large point clouds with 1000 or more points are needed for the more complex block oligomers. The trained PointNet model achieves an accuracy as high as 0.99 for globally ordered morphologies formed by linear diblock, linear triblock, and 3-arm and 4-arm star-block oligomers, and it also allows for the discovery of emerging ordered patterns from nonequilibrium systems.

Machine learning for condensed matter physics

Condensed matter physics (CMP) seeks to understand the microscopic interactions of matter at the quantum and atomistic levels, and describes how these interactions result in both mesoscopic and macroscopic properties. CMP overlaps with many other important branches of science, such as chemistry, materials science, statistical physics, and high-performance computing. With the advancements in modern machine learning (ML) technology, a keen interest in applying these algorithms to further CMP research has created a compelling new area of research at the intersection of both fields. In this review, we aim to explore the main areas within CMP, which have successfully applied ML techniques to further research, such as the description and use of ML schemes for potential energy surfaces, the characterization of topological phases of matter in lattice systems, the prediction of phase transitions in off-lattice and atomistic simulations, the interpretation of ML theories with physics-inspired frameworks and the enhancement of simulation methods with ML algorithms. We also discuss in detail the main challenges and drawbacks of using ML methods on CMP problems, as well as some perspectives for future developments.

Integrated Wound Recognition in Bandages for Intelligent Treatment

The mismatch between irregular wounds and fixed dressings is an important but long-neglected problem in wound management, which may cause imprecise or incomplete care outcomes. Continuous advances in the internet, Internet of Things, and materials science put forward higher requirements in precise wound treatment. Herein, an integrated wound recognition strategy is demonstrated by extracting digital geometric information of specific irregular wounds. Such online wound image scanning and recognition integrated with offline intelligent material fabrication can greatly reduce incomplete wound closure or potential irritation toward normal tissue around the wound site. As a proof-of-concept, an active matrix of Ag-modified gelatin on bandages demonstrates promising antibacterial properties for practical application. In vivo mouse experiment reveals that wound-customized bands enable precisely fitting in the irregular wound bed, resulting in accelerating wound healing efficiency. Such wound recognitions in a bandage involve multidisciplinary technology including image identification, computer modeling nanomaterial fabrication and modification, presenting a broad potential of personalized wound healthcare for a wide variety of clinical applications.

Multivalent Surface Cations Enhance Heterogeneous Freezing of Water on Muscovite Mica

Heterogeneous ice nucleation is a crucial phenomenon in various fields of fundamental and applied science. We investigate the effect of surface cations on freezing of water on muscovite mica. Mica is unique in that the exposed ion on its surface can be readily and easily exchanged without affecting other properties such as surface roughness. We investigate freezing on natural (K⁺) mica and mica in which we have exchanged K⁺ for Al³⁺, Mg²⁺, Ca²⁺, and Sr²⁺. We find that liquid water freezes at higher temperatures when ions of higher valency are present on the surface, thus exposing more of the underlying silica layer. Our data also show that the size of the ion affects the characteristic freezing temperature. Using molecular dynamics simulations, we investigate the effects that the ion valency and exposed silica layer have on the behavior of water on the surface. The results indicate that multivalent cations enhance the probability of forming large clusters of hydrogen bonded water molecules that are anchored by the hydration shells of the cations. These clusters also have a large fraction of free water that can reorient to take ice-like configurations, which are promoted by the regions on mica devoid of the ions. Thus, these clusters could serve as seedbeds for ice nuclei. The combined experimental and simulation studies shed new light on the influence of surface ions on heterogeneous ice nucleation.

Big-Data Science in Porous Materials: Materials Genomics and Machine Learning

By combining metal nodes with organic linkers we can potentially synthesize millions of possible metal-organic frameworks (MOFs). The fact that we have so many materials opens many exciting avenues but also create new challenges. We simply have too many materials to be processed using conventional, brute force, methods. In this review, we show that having so many materials allows us to use big-data methods as a powerful technique to study these materials and to discover complex correlations. The first part of the review gives an introduction to the principles of big-data science. We show how to select appropriate training sets, survey approaches that are used to represent these materials in feature space, and review different learning architectures, as well as evaluation and interpretation strategies. In the second part, we review how the different approaches of machine learning have been applied to porous materials. In particular, we discuss applications in the field of gas storage and separation, the stability of these materials, their electronic properties, and their synthesis. Given the increasing interest of the scientific community in machine learning, we expect this list to rapidly expand in the coming years.

GCIceNet: a graph convolutional network for accurate classification of water phases

We develop GCIceNet, which automatically generates machine-based order parameters for classifying the phases of water molecules via supervised and unsupervised learning with graph convolutional networks.