GradNav: Accelerated Exploration of Potential Energy Surfaces with Gradient-Based Navigation.

Exploring the potential energy surface (PES) of molecular systems is important for comprehending their complex behaviors, particularly through the identification of various metastable states. However, the transition between these states is often hindered by substantial energy barriers, demanding prolonged molecular simulations that consume considerable computational resources. Our study introduces the gradient-based navigation (GradNav) algorithm, which accelerates the exploration of the energy surface and enables proper reconstruction of the PES. This algorithm employs a strategy of initiating short simulation runs from updated starting points derived from prior observations to effectively navigate across potential barriers and explore new regions. To evaluate GradNav's performance, we introduce two metrics: the deepest well escape frame (DWEF) and the search success initialization ratio (SSIR). Through applications on Langevin dynamics within Müller-type PESs and molecular dynamics simulations of the Fs-peptide protein, these metrics demonstrate GradNav's enhanced ability to escape deep energy wells and its reduced reliance on initial conditions, as denoted by the reduced DWEF values and increased SSIR values, respectively. Consequently, this improved exploration capability enables more precise energy estimations from simulation trajectories.

Fast Water Desalination with a Graphene-MoS2 Nanoporous Heterostructure.

Energy-efficient water desalination is the key to tackle the challenges with drought and water scarcity that affect 1.2 billion people. The material and type of membrane in reverse osmosis water desalination are the key factors in their efficiency. In this work, we explored the potential of a graphene-MoS2 heterostructure membrane for water desalination, focusing on bilayer membranes and their advantages over monolayer counterparts. Through extensive molecular dynamics simulation and statistical analysis, the bilayer MoS2-graphene was investigated and compared to the monolayer of graphene and MoS2. By optimizing the heterostructure membrane, improved water flux was achieved while maintaining a high ion rejection rate. Furthermore, the study delves into the physical mechanisms underlying the superior performance of heterostructure nanopores, comparing them with circular bilayer and monolayer pores. Factors investigated include water structure, hydration shell near the membrane surface, water density, energy barrier using the potential of mean force, and porosity within the nanopore. Our findings contribute to the understanding of heterostructure membranes and their potential in enhancing the water desalination efficiency, providing valuable insights for future membrane design and optimization.

Machine Learning in Membrane Design: From Property Prediction to AI-Guided Optimization.

Porous membranes, either polymeric or two-dimensional materials, have been extensively studied because of their outstanding performance in many applications such as water filtration. Recently, inspired by the significant success of machine learning (ML) in many areas of scientific discovery, researchers have started to tackle the problem in the field of membrane design using data-driven ML tools. In this Mini Review, we summarize research efforts on three types of applications of machine learning in membrane design, including (1) membrane property prediction using ML, (2) gaining physical insight and drawing quantitative relationships between membrane properties and performance using explainable artificial intelligence, and (3) ML-guided design, optimization, or virtual screening of membranes. On top of the review of previous research, we discuss the challenges associated with applying ML for membrane design and potential future directions.

Pretraining Strategies for Structure Agnostic Material Property Prediction.

In recent years, machine learning (ML), especially graph neural network (GNN) models, has been successfully used for fast and accurate prediction of material properties. However, most ML models rely on relaxed crystal structures to develop descriptors for accurate predictions. Generating these relaxed crystal structures can be expensive and time-consuming, thus requiring an additional processing step for models that rely on them. To address this challenge, structure-agnostic methods have been developed, which use fixed-length descriptors engineered based on human knowledge about the material. However, the fixed-length descriptors are often hand-engineered and require extensive domain knowledge and generally are not used in the context of learnable models which are known to have a superior performance. Recent advancements have proposed learnable frameworks that can construct representations based on stoichiometry alone, allowing the flexibility of using deep learning frameworks as well as leveraging structure-agnostic learning. In this work, we propose three different pretraining strategies that can be used to pretrain these structure-agnostic, learnable frameworks to further improve the downstream material property prediction performance. We incorporate strategies such as self-supervised learning (SSL), fingerprint learning (FL), and multimodal learning (ML) and demonstrate their efficacy on downstream tasks for the Roost architecture, a popular structure-agnostic framework. Our results show significant improvement in small data sets and data efficiency in the larger data sets, underscoring the potential of our pretrain strategies that effectively leverage unlabeled data for accurate material property prediction.

GPCR-BERT: Interpreting Sequential Design of G Protein-Coupled Receptors Using Protein Language Models.

With the rise of transformers and large language models (LLMs) in chemistry and biology, new avenues for the design and understanding of therapeutics have been opened up to the scientific community. Protein sequences can be modeled as language and can take advantage of recent advances in LLMs, specifically with the abundance of our access to the protein sequence data sets. In this letter, we developed the GPCR-BERT model for understanding the sequential design of G protein-coupled receptors (GPCRs). GPCRs are the target of over one-third of Food and Drug Administration-approved pharmaceuticals. However, there is a lack of comprehensive understanding regarding the relationship among amino acid sequence, ligand selectivity, and conformational motifs (such as NPxxY, CWxP, and E/DRY). By utilizing the pretrained protein model (Prot-Bert) and fine-tuning with prediction tasks of variations in the motifs, we were able to shed light on several relationships between residues in the binding pocket and some of the conserved motifs. To achieve this, we took advantage of attention weights and hidden states of the model that are interpreted to extract the extent of contributions of amino acids in dictating the type of masked ones. The fine-tuned models demonstrated high accuracy in predicting hidden residues within the motifs. In addition, the analysis of embedding was performed over 3D structures to elucidate the higher-order interactions within the conformations of the receptors.

PeptideBERT: A Language Model Based on Transformers for Peptide Property Prediction.

Recent advances in language models have enabled the protein modeling community with a powerful tool that uses transformers to represent protein sequences as text. This breakthrough enables a sequence-to-property prediction for peptides without relying on explicit structural data. Inspired by the recent progress in the field of large language models, we present PeptideBERT, a protein language model specifically tailored for predicting essential peptide properties such as hemolysis, solubility, and nonfouling. The PeptideBERT utilizes the ProtBERT pretrained transformer model with 12 attention heads and 12 hidden layers. Through fine-tuning the pretrained model for the three downstream tasks, our model is state of the art (SOTA) in predicting hemolysis, which is crucial for determining a peptide's potential to induce red blood cells as well as nonfouling properties. Leveraging primarily shorter sequences and a data set with negative samples predominantly associated with insoluble peptides, our model showcases remarkable performance.

Unveiling Switching Function of Amino Acids in Proteins Using a Machine Learning Approach.

Dynamics of individual amino acids play key roles in the overall properties of proteins. However, the knowledge of protein structural features at the residue level is limited due to the current resolutions of experimental and computational techniques. To address this issue, we designed a novel machine-learning (ML) framework that uses Molecular Dynamics (MD) trajectories to identify the major conformational states of individual amino acids, classify amino acids switching between two distinct modes, and evaluate their degree of dynamic stability. The Random Forest model achieved 96.94% classification accuracy in identifying switch residues within proteins. Additionally, our framework distinguishes between the stable switch (SS) residues, which remain stable in one angular state and jump once to another state during protein dynamics, and unstable switch (US) residues, which constantly fluctuate between the two angular states. This study also illustrates the correlation between the dynamics of SS residues and the protein's global properties.

Neural network predicts ion concentration profiles under nanoconfinement.

Modeling the ion concentration profile in nanochannel plays an important role in understanding the electrical double layer and electro-osmotic flow. Due to the non-negligible surface interaction and the effect of discrete solvent molecules, molecular dynamics (MD) simulation is often used as an essential tool to study the behavior of ions under nanoconfinement. Despite the accuracy of MD simulation in modeling nanoconfinement systems, it is computationally expensive. In this work, we propose neural network to predict ion concentration profiles in nanochannels with different configurations, including channel widths, ion molarity, and ion types. By modeling the ion concentration profile as a probability distribution, our neural network can serve as a much faster surrogate model for MD simulation with high accuracy. We further demonstrate the superior prediction accuracy of neural network over XGBoost. Finally, we demonstrated that neural network is flexible in predicting ion concentration profiles with different bin sizes. Overall, our deep learning model is a fast, flexible, and accurate surrogate model to predict ion concentration profiles in nanoconfinement.

Denoise Pretraining on Nonequilibrium Molecules for Accurate and Transferable Neural Potentials.

Recent advances in equivariant graph neural networks (GNNs) have made deep learning amenable to developing fast surrogate models to expensive ab initio quantum mechanics (QM) approaches for molecular potential predictions. However, building accurate and transferable potential models using GNNs remains challenging, as the data are greatly limited by the expensive computational costs and level of theory of QM methods, especially for large and complex molecular systems. In this work, we propose denoise pretraining on nonequilibrium molecular conformations to achieve more accurate and transferable GNN potential predictions. Specifically, atomic coordinates of sampled nonequilibrium conformations are perturbed by random noises, and GNNs are pretrained to denoise the perturbed molecular conformations which recovers the original coordinates. Rigorous experiments on multiple benchmarks reveal that pretraining significantly improves the accuracy of neural potentials. Furthermore, we show that the proposed pretraining approach is model-agnostic, as it improves the performance of different invariant and equivariant GNNs. Notably, our models pretrained on small molecules demonstrate remarkable transferability, improving performance when fine-tuned on diverse molecular systems, including different elements, charged molecules, biomolecules, and larger systems. These results highlight the potential for leveraging denoise pretraining approaches to build more generalizable neural potentials for complex molecular systems.

Activity Map and Transition Pathways of G Protein-Coupled Receptor Revealed by Machine Learning.

Approximately, one-third of all U.S. Food and Drug Administration approved drugs target G protein-coupled receptors (GPCRs). However, more knowledge of protein structure-activity correlation is required to improve the efficacy of the drugs targeting GPCRs. In this study, we developed a machine learning model to predict the activation state and activity level of the receptors with high prediction accuracy. Furthermore, we applied this model to thousands of molecular dynamics trajectories to correlate residue-level conformational changes of a GPCR to its activity level. Finally, the most probable transition pathway between activation states of a receptor can be identified using the state-activity information. In addition, with this model, we can associate the contribution of each amino acid to the activation process. Using this method, we can design drugs that mainly target principal amino acids driving the transition between activation states of GPCRs. Our advanced method is generalizable to all GPCR classes and provides mechanistic insight into the activation mechanism in the receptors.

MOFormer: Self-Supervised Transformer Model for Metal-Organic Framework Property Prediction.

Metal-organic frameworks (MOFs) are materials with a high degree of porosity that can be used for many applications. However, the chemical space of MOFs is enormous due to the large variety of possible combinations of building blocks and topology. Discovering the optimal MOFs for specific applications requires an efficient and accurate search over countless potential candidates. Previous high-throughput screening methods using computational simulations like DFT can be time-consuming. Such methods also require the 3D atomic structures of MOFs, which adds one extra step when evaluating hypothetical MOFs. In this work, we propose a structure-agnostic deep learning method based on the Transformer model, named as MOFormer, for property predictions of MOFs. MOFormer takes a text string representation of MOF (MOFid) as input, thus circumventing the need of obtaining the 3D structure of a hypothetical MOF and accelerating the screening process. By comparing to other descriptors such as Stoichiometric-120 and revised autocorrelations, we demonstrate that MOFormer can achieve state-of-the-art structure-agnostic prediction accuracy on all benchmarks. Furthermore, we introduce a self-supervised learning framework that pretrains the MOFormer via maximizing the cross-correlation between its structure-agnostic representations and structure-based representations of the crystal graph convolutional neural network (CGCNN) on >400k publicly available MOF data. Benchmarks show that pretraining improves the prediction accuracy of both models on various downstream prediction tasks. Furthermore, we revealed that MOFormer can be more data-efficient on quantum-chemical property prediction than structure-based CGCNN when training data is limited. Overall, MOFormer provides a novel perspective on efficient MOF property prediction using deep learning.

The diffusion mechanism of water in conductive metal-organic frameworks.

Water is one of the most important guest molecules in metal-organic frameworks (MOFs) since it often serves as a solvent for ions and other molecules. Studying the diffusion mechanism of water molecules in conductive MOFs (c-MOFs) is fundamental to harnessing the potential of c-MOFs in designing next generation energy storage devices. In this work, using molecular dynamics simulations, we show that water follows the Fickian-type of diffusion mechanism in different types of c-MOFs. We investigate the effect of the stacking and metal center type on the water diffusion coefficient in c-MOFs. Water in c-MOFs with eclipsed stacking is shown to have 21.5% higher diffusion coefficient than in c-MOFs with slipped-parallel stacking, and 4-8% higher diffusion coefficient than in bulk water. The physical reasons behind the reduced water diffusion coefficient in slipped-parallel stacking c-MOFs are the higher number of hydrogen bonds near the inner surface and the zig-zag geometry. This work provides a molecular insight into the water dynamics and water structure inside multiple types of c-MOFs.

Which 2D Material is Better for DNA Detection: Graphene, MoS2, or MXene?

Despite much research on characterizing 2D materials for DNA detection with nanopore technology, a thorough comparison between the performance of different 2D materials is currently lacking. In this work, using extensive molecular dynamics simulations, we compare nanoporous graphene, MoS2 and titanium carbide MXene (Ti3C2) for their DNA detection performance and sensitivity. The ionic current and residence time of DNA are characterized in each nanoporous materials by performing hundreds of simulations. We devised two statistical measures including the Kolmogorov-Smirnov test and the absolute pairwise difference to compare the performance of nanopores. We found that graphene nanopore is the most sensitive membrane for distinguishing DNA bases. The MoS2 is capable of distinguishing the A and T bases from the C and G bases better than graphene and MXene. Physisorption and the orientation of DNA in nanopores are further investigated to provide molecular insight into the performance characteristics of different nanopores.

Engineering rotating apical-out airway organoid for assessing respiratory cilia motility.

Motile cilia project from the airway apical surface and directly interface with inhaled external environment. Owing to cilia's nanoscale dimension and high beating frequency, quantitative assessment of their motility remains a sophisticated task. Here we described a robust approach for reproducible engineering of apical-out airway organoid (AOAO) from a defined number of cells. Propelled by exterior-facing cilia beating, the mature AOAO exhibited stable rotational motion when surrounded by Matrigel. We developed a computational framework leveraging computer vision algorithms to quantify AOAO rotation and correlated it with the direct measurement of cilia motility. We further established the feasibility of using AOAO rotation to recapitulate and measure defective cilia motility caused by chemotherapy-induced toxicity and by CCDC39 mutations in cells from patients with primary ciliary dyskinesia. We expect our rotating AOAO model and the associated computational pipeline to offer a generalizable framework to expedite the modeling of and therapeutic development for genetic and environmental ciliopathies.

Prediction of GPCR activity using machine learning.

GPCRs are the target for one-third of the FDA-approved drugs, however; the development of new drug molecules targeting GPCRs is limited by the lack of mechanistic understanding of the GPCR structure-activity-function relationship. To modulate the GPCR activity with highly specific drugs and minimal side-effects, it is necessary to quantitatively describe the important structural features in the GPCR and correlate them to the activation state of GPCR. In this study, we developed 3 ML approaches to predict the conformation state of GPCR proteins. Additionally, we predict the activity level of GPCRs based on their structure. We leverage the unique advantages of each of the 3 ML approaches, interpretability of XGBoost, minimal feature engineering for 3D convolutional neural network, and graph representation of protein structure for graph neural network. By using these ML approaches, we are able to predict the activation state of GPCRs with high accuracy (91%-95%) and also predict the activation state of GPCRs with low error (MAE of 7.15-10.58). Furthermore, the interpretation of the ML approaches allows us to determine the importance of each of the features in distinguishing between the GPCRs conformations.

Improving Molecular Contrastive Learning via Faulty Negative Mitigation and Decomposed Fragment Contrast.

Deep learning has been a prevalence in computational chemistry and widely implemented in molecular property predictions. Recently, self-supervised learning (SSL), especially contrastive learning (CL), has gathered growing attention for the potential to learn molecular representations that generalize to the gigantic chemical space. Unlike supervised learning, SSL can directly leverage large unlabeled data, which greatly reduces the effort to acquire molecular property labels through costly and time-consuming simulations or experiments. However, most molecular SSL methods borrow the insights from the machine learning community but neglect the unique cheminformatics (e.g., molecular fingerprints) and multilevel graphical structures (e.g., functional groups) of molecules. In this work, we propose iMolCLR, improvement of Molecular Contrastive Learning of Representations with graph neural networks (GNNs) in two aspects: (1) mitigating faulty negative contrastive instances via considering cheminformatics similarities between molecule pairs and (2) fragment-level contrasting between intramolecule and intermolecule substructures decomposed from molecules. Experiments have shown that the proposed strategies significantly improve the performance of GNN models on various challenging molecular property predictions. In comparison to the previous CL framework, iMolCLR demonstrates an averaged 1.2% improvement of ROC-AUC on eight classification benchmarks and an averaged 10.1% decrease of the error on six regression benchmarks. On most benchmarks, the generic GNN pretrained by iMolCLR rivals or even surpasses supervised learning models with sophisticated architectures and engineered features. Further investigations demonstrate that representations learned through iMolCLR intrinsically embed scaffolds and functional groups that can reason molecule similarities.

Graph neural networks accelerated molecular dynamics.

Molecular Dynamics (MD) simulation is a powerful tool for understanding the dynamics and structure of matter. Since the resolution of MD is atomic-scale, achieving long timescale simulations with femtosecond integration is very expensive. In each MD step, numerous iterative computations are performed to calculate energy based on different types of interaction and their corresponding spatial gradients. These repetitive computations can be learned and surrogated by a deep learning model, such as a Graph Neural Network (GNN). In this work, we developed a GNN Accelerated MD (GAMD) model that directly predicts forces, given the state of the system (atom positions, atom types), bypassing the evaluation of potential energy. By training the GNN on a variety of data sources (simulation data derived from classical MD and density functional theory), we show that GAMD can predict the dynamics of two typical molecular systems, Lennard-Jones system and water system, in the NVT ensemble with velocities regulated by a thermostat. We further show that GAMD's learning and inference are agnostic to the scale, where it can scale to much larger systems at test time. We also perform a comprehensive benchmark test comparing our implementation of GAMD to production-level MD software, showing GAMD's competitive performance on the large-scale simulation.

Activation Pathways of Neurotensin Receptor 1 Elucidated Using Statistical Machine Learning.

Neurotensin receptor 1 (NTSR1) is a G-protein coupled receptor (GPCR) that mediates many biological processes through its interaction with the neurotensin (NTS) peptide. The NTSR1 protein is a clinically significant target as it is involved in the proliferation of cancer cells. Understanding the activation mechanism of NTSR1 is an important prerequisite for exploring the therapeutic potential of targeting NTSR1 and the development of drug molecules specific to NTSR1. Previous studies have been aimed at elucidating the structure of NTSR1 in the active and inactive conformations; however, the intermediate molecular pathway for NTSR1 activation dynamics is largely unknown. In this study, we performed extensive molecular dynamics (MD) simulations of the NTSR1 protein and analyzed its kinetic conformational changes to determine the microswitches that drive NTSR1 activation. To biophysically interpret the high-dimensional simulation trajectories, we used Markov state models and machine learning to elucidate the important and detailed conformational changes in NTSR1. Through the analysis of identified microswitches, we propose a mechanistic pathway for NTSR1 activation.

Forecasting COVID-19 new cases using deep learning methods.

After nearly two years since the first identification of SARS-CoV-2 virus, the surge in cases because of virus mutations is a cause of grave public health concern across the globe. As a result of this health crisis, predicting the transmission pattern of the virus is one of the most vital tasks for preparing and controlling the pandemic. In addition to mathematical models, machine learning tools, especially deep learning models have been developed for forecasting the trend of the number of patients affected by SARS-CoV-2 with great success. In this paper, three deep learning models, including CNN, LSTM, and the CNN-LSTM have been developed to predict the number of COVID-19 cases for Brazil, India and Russia. We also compare the performance of our models with the previously developed deep learning models and notice significant improvements in prediction performance. Although our models have been used only for forecasting cases in these three countries, the models can be easily applied to datasets of other countries. Among the models developed in this work, the LSTM model has the highest performance when forecasting and shows an improvement in the forecasting accuracy compared with some existing models. The research will enable accurate forecasting of the COVID-19 cases and support the global fight against the pandemic.

Understanding mutation hotspots for the SARS-CoV-2 spike protein using Shannon Entropy and K-means clustering.

The SARS-CoV-2 virus like many other viruses has transformed in a continual manner to give rise to new variants by means of mutations commonly through substitutions and indels. These mutations in some cases can give the virus a survival advantage making the mutants dangerous. In general, laboratory investigation must be carried to determine whether the new variants have any characteristics that can make them more lethal and contagious. Therefore, complex and time-consuming analyses are required in order to delve deeper into the exact impact of a particular mutation. The time required for these analyses makes it difficult to understand the variants of concern and thereby limiting the preventive action that can be taken against them spreading rapidly. In this analysis, we have deployed a statistical technique Shannon Entropy, to identify positions in the spike protein of SARS Cov-2 viral sequence which are most susceptible to mutations. Subsequently, we also use machine learning based clustering techniques to cluster known dangerous mutations based on similarities in properties. This work utilizes embeddings generated using language modeling, the ProtBERT model, to identify mutations of a similar nature and to pick out regions of interest based on proneness to change. Our entropy-based analysis successfully predicted the fifteen hotspot regions, among which we were able to validate ten known variants of interest, in six hotspot regions. As the situation of SARS-COV-2 virus rapidly evolves we believe that the remaining nine mutational hotspots may contain variants that can emerge in the future. We believe that this may be promising in helping the research community to devise therapeutics based on probable new mutation zones in the viral sequence and resemblance in properties of various mutations.