Morgan Fingerprints Research Articles

Prediction of potential antitumor components in Ganoderma lucidum: A combined approach using machine learning and molecular docking

The objective of this study is to develop a reliable predictive model for antitumour activity and to identify potential antitumour components in Ganoderma lucidum. Four machine learning models, including Random Forest, were employed to train predictive models for antitumour activity, utilising Morgan fingerprints as molecular descriptors. The most effective model was then employed to predict the chemical composition of Ganoderma lucidum, identifying the four most probable compounds for molecular docking with known TNF-α-related targets. The findings of the study indicate that a Support Vector Machine (SVM) model exhibits an accuracy, F1 score, AUC, and sensitivity of 0.7638, 0.7638, 0.8332, and 0.7621, respectively. The model demonstrated an 80 % accuracy rate in predicting the antitumour activity of 10 FDA-approved drugs. Besides, the model identified 11 components in Ganoderma lucidum, including 3-nitroanisole, with a probability of antitumour activity exceeding 0.5, indicating their potential as antitumour agents. The results of the molecular docking procedure indicated that the four most promising antitumour compounds derived from Ganoderma lucidum exhibited a favourable binding affinity with the TNF-α target. In conclusion, this study incorporated a machine learning prediction step prior to molecular docking, thereby enhancing the reliability of the latter. Furthermore, it identified previously unreported compounds in Ganoderma lucidum with potential antitumour activity, such as 3-nitroanisole.

StructNet-DDI: Molecular Structure Characterization-Based ResNet for Prediction of Drug-Drug Interactions.

This study introduces a deep learning framework based on SMILES representations of chemical structures to predict drug-drug interactions (DDIs). The model extracts Morgan fingerprints and key molecular descriptors, transforming them into raw graphical features for input into a modified ResNet18 architecture. The deep residual network, enhanced with regularization techniques, efficiently addresses training issues such as gradient vanishing and exploding, resulting in superior predictive performance. Experimental results show that StructNet-DDI achieved an AUC of 99.7%, an accuracy of 94.4%, and an AUPR of 99.9%, demonstrating the model's effectiveness and reliability. These findings highlight that StructNet-DDI can effectively extract crucial features from molecular structures, offering a simple yet robust tool for DDI prediction.

Multi-property prediction and high-throughput screening of polyimides: An application case for interpretable machine learning

Polyimide (PI), as a high-performance polymer widely used in aerospace, optoelectronics, microelectronics, etc., the properties it focused on in different areas of application were diverse. However, most of the past machine learning studies in polyimide property prediction were focused only on the prediction of a single property. This study focused on four major categories of properties of polyimide, including thermal (Tg, Td5, Td10, and CTE), mechanical (Ts and TM), dielectric (ε), and optical (λcutoff, T400, nav, and Δn), totaling eleven properties. PI data synthesized from previous studies were collected. MorGan fingerprints, improved MorGan fingerprints, RDkit and Mordred descriptors were selected as feature representations. Four ML models such as DNN, RF, XGBoost and BT were also built. Collectively, 176 machine learning models were trained for 11 predictions of properties. The performance and generalization ability of the models were confirmed by experimental validation, external validation and leave-one-out cross-validation. SHAP analysis was used to explain the optimal model for each property prediction from a physicochemical point of view and structural aspects, and three PIs with different structures were designed accordingly. Finally, a high-throughput virtual screening of nearly 7.6 million polyimides was performed based on the trained model. SA_scores was used to evaluate the ease of synthesis of each PI, and finally high-performance PIs with potential and easy to synthesize were selected for each field. This study could be expected to provide a guideline and a design framework for the application of PIs in various fields in the future.

Accelerating the discovery of type Ⅱ photosensitizer: Experimentally validated machine learning models for predicting the singlet oxygen quantum yield of photosensitive molecule

Photodynamic therapy (PDT) is an emerging cancer treatment that mainly relies on photosensitizer (PS) to generate singlet oxygen for tumor destruction. Developing PSs with high singlet oxygen quantum yields (SO-QYs) requires extensive experimentation, limiting their rapid screening. Herein, to streamline this process, this study introduces several machine learning (ML) models that accurately predicts SO-QY across various experimental conditions. The models’ establishment is based on two feature matrices derived from Morgan fingerprints (MFPs) and descriptors (molecular descriptors and quantum chemical descriptors, MD_QCDs). Comparative and evaluative results indicate that the XGBoost model constructed with MFPs and the AdaBoost model constructed with MD_QCDs exhibit superior predictive performance, with R2 values of 0.8648 and 0.8460, respectively. Furthermore, by utilizing SHapley Additive exPlanations (SHAP) analysis and quantum chemistry, we analyzed that the iodine atoms and larger conjugated systems significantly influenced the SO-QY. Experimental validation, based on this analysis, demonstrates that our models not only possess excellent predictive capabilities but also exhibit strong interpretability. In summary, this work has established several interpretable models with outstanding predictive performance, which can aid in the more rapid screening of PSs, thus promoting their application in PDT.

MMFSyn: A Multimodal Deep Learning Model for Predicting Anticancer Synergistic Drug Combination Effect.

Combination therapy aims to synergistically enhance efficacy or reduce toxic side effects and has widely been used in clinical practice. However, with the rapid increase in the types of drug combinations, identifying the synergistic relationships between drugs remains a highly challenging task. This paper proposes a novel deep learning model MMFSyn based on multimodal drug data combined with cell line features. Firstly, to ensure the full expression of drug molecular features, multiple modalities of drugs, including Morgan fingerprints, atom sequences, molecular diagrams, and atomic point cloud data, are extracted using SMILES. Secondly, for different modal data, a Bi-LSTM, gMLP, multi-head attention mechanism, and multi-scale GCNs are comprehensively applied to extract the drug feature. Then, it selects appropriate omics features from gene expression and mutation omics data of cancer cell lines to construct cancer cell line features. Finally, these features are combined to predict the synergistic anti-cancer drug combination effect. The experimental results verify that MMFSyn has significant advantages in performance compared to other popular methods, with a root mean square error of 13.33 and a Pearson correlation coefficient of 0.81, which indicates that MMFSyn can better capture the complex relationship between multimodal drug combinations and omics data, thereby improving the synergistic drug combination prediction.

Cheminformatics analysis of indoleamine and tryptophan 2,3-dioxygenase inhibitors: A descriptor and fingerprint based machine learning approach to disclose selectivity measures

Indoleamine 2,3-dioxygenase (IDO) and tryptophan 2,3-dioxygenase (TDO) are attractive drug targets for cancer immunotherapy. After disappointing results of the epacadostat as a selective IDO inhibitor in phase III clinical trials, there is much interest in the development of the TDO selective inhibitors. In the current study, several data analysis methods and machine learning approaches including logistic regression, Random Forest, XGBoost and Support Vector Machines were used to model a data set of compounds retrieved from ChEMBL. Models based on the Morgan fingerprints revealed notable fragments for the selective inhibition of the IDO, TDO or both. Multiple fragment docking was performed to find the best set of bound fragments and their orientation in the space for efficient linking. Linking the fragments and optimization of the final molecules were accomplished by means of an artificial intelligence generative framework. Finally, selectivity of the optimized molecules was assessed and the top 4 lead molecules were filtered through PAINS, Brenk and NIH filters. Results indicated that phenyloxalamide, fluoroquinoline, and 3-bromo-4-fluroaniline confer selectivity towards the IDO inhibition. Correspondingly, 1-benzyl-1H-naphtho[2,3-d][1,2,3]triazole-4,9-dione was found to be an integral fragment for the selective inhibition of the TDO by constituting a coordination bond with the Fe atom of heme. In addition, furo[2,3-c]pyridine-2,3-diamine was found as a common fragment for inhibition of the both targets and can be used in the design of the dual target inhibitors of the IDO and TDO. The new fragments introduced here can be a useful building blocks for incorporation into the selective TDO or dual IDO/TDO inhibitors.

Drug-Target Interaction Prediction Based on an Interactive Inference Network.

Drug-target interactions underlie the actions of chemical substances in medicine. Moreover, drug repurposing can expand use profiles while reducing costs and development time by exploiting potential multi-functional pharmacological properties based upon additional target interactions. Nonetheless, drug repurposing relies on the accurate identification and validation of drug-target interactions (DTIs). In this study, a novel drug-target interaction prediction model was developed. The model, based on an interactive inference network, contains embedding, encoding, interaction, feature extraction, and output layers. In addition, this study used Morgan and PubChem molecular fingerprints as additional information for drug encoding. The interaction layer in our model simulates the drug-target interaction process, which assists in understanding the interaction by representing the interaction space. Our method achieves high levels of predictive performance, as well as interpretability of drug-target interactions. Additionally, we predicted and validated 22 Alzheimer's disease-related targets, suggesting our model is robust and effective and thus may be beneficial for drug repurposing.

Predicting blood–brain barrier permeability of molecules with a large language model and machine learning

Predicting the blood–brain barrier (BBB) permeability of small-molecule compounds using a novel artificial intelligence platform is necessary for drug discovery. Machine learning and a large language model on artificial intelligence (AI) tools improve the accuracy and shorten the time for new drug development. The primary goal of this research is to develop artificial intelligence (AI) computing models and novel deep learning architectures capable of predicting whether molecules can permeate the human blood–brain barrier (BBB). The in silico (computational) and in vitro (experimental) results were validated by the Natural Products Research Laboratories (NPRL) at China Medical University Hospital (CMUH). The transformer-based MegaMolBART was used as the simplified molecular input line entry system (SMILES) encoder with an XGBoost classifier as an in silico method to check if a molecule could cross through the BBB. We used Morgan or Circular fingerprints to apply the Morgan algorithm to a set of atomic invariants as a baseline encoder also with an XGBoost classifier to compare the results. BBB permeability was assessed in vitro using three-dimensional (3D) human BBB spheroids (human brain microvascular endothelial cells, brain vascular pericytes, and astrocytes). Using multiple BBB databases, the results of the final in silico transformer and XGBoost model achieved an area under the receiver operating characteristic curve of 0.88 on the held-out test dataset. Temozolomide (TMZ) and 21 randomly selected BBB permeable compounds (Pred scores = 1, indicating BBB-permeable) from the NPRL penetrated human BBB spheroid cells. No evidence suggests that ferulic acid or five BBB-impermeable compounds (Pred scores < 1.29423E−05, which designate compounds that pass through the human BBB) can pass through the spheroid cells of the BBB. Our validation of in vitro experiments indicated that the in silico prediction of small-molecule permeation in the BBB model is accurate. Transformer-based models like MegaMolBART, leveraging the SMILES representations of molecules, show great promise for applications in new drug discovery. These models have the potential to accelerate the development of novel targeted treatments for disorders of the central nervous system.

Thermal stability prediction of copolymerized polyimides via an interpretable transfer learning model

To address the issues with molecular representation of copolymerized polyimides (PIs) and the mini dataset of PI powders. We constructed an interpretable machine learning (ML) model for PI films using the weighted-additive Morgan Fingerprints with Frequency descriptors and developed an interpretable transfer learning model for PI powders. To enhance Thermal Stability (Temperature at 5% weight loss) of PI films and powders, it is recommended to add conjugated functional groups to diamines, control phenyl ring side chains, and reduce pyridine and hydroxyl groups; select copolyimides (co-PIs); ensure that anhydride is directly connected to the benzene ring in dianhydrides, avoiding aliphatic cycles. It is noteworthy that the close alignment between experimental results and model predictions serves to confirm the model is a reliable prediction tool. It is hoped that this polymer informatics approach will provide further implementation for practical applications of other functional materials.

Incorporating Synthetic Accessibility in Drug Design: Predicting Reaction Yields of Suzuki Cross-Couplings by Leveraging AbbVie's 15-Year Parallel Library Data Set.

Despite the increased use of computational tools to supplement medicinal chemists' expertise and intuition in drug design, predicting synthetic yields in medicinal chemistry endeavors remains an unsolved challenge. Existing design workflows could profoundly benefit from reaction yield prediction, as precious material waste could be reduced, and a greater number of relevant compounds could be delivered to advance the design, make, test, analyze (DMTA) cycle. In this work, we detail the evaluation of AbbVie's medicinal chemistry library data set to build machine learning models for the prediction of Suzuki coupling reaction yields. The combination of density functional theory (DFT)-derived features and Morgan fingerprints was identified to perform better than one-hot encoded baseline modeling, furnishing encouraging results. Overall, we observe modest generalization to unseen reactant structures within the 15-year retrospective library data set. Additionally, we compare predictions made by the model to those made by expert medicinal chemists, finding that the model can often predict both reaction success and reaction yields with greater accuracy. Finally, we demonstrate the application of this approach to suggest structurally and electronically similar building blocks to replace those predicted or observed to be unsuccessful prior to or after synthesis, respectively. The yield prediction model was used to select similar monomers predicted to have higher yields, resulting in greater synthesis efficiency of relevant drug-like molecules.

Alternative weighting schemes for fine‐tuned extended similarity indices

AbstractExtended similarity indices (i.e., generalization of pairwise similarity) have recently gained importance because of their simplicity, fast computation, and superiority in tasks like diversity picking. However, they operate with several meta parameters that should be optimized. Earlier, we extended the binary similarity indices to “discrete non‐binary” and “continuous” data; now we continue with introducing and comparing multiple weighting functions. As a case study, the similarity of CYP enzyme inhibitors (4016 molecules after curation) was characterized by their extended similarities, based on 2D descriptors, MACCS and Morgan fingerprints. A statistical workflow based on sum of ranking differences (SRD) and analysis of variance (ANOVA) was used for finding the optimal weight function(s). Overall, the best weighting function is the fraction (“frac”), which corresponds to the principle of parsimony. Optimal extended similarity indices were also found, and their differences are revealed across different data sets. We intend this work to be a guideline for users of extended similarity indices regarding the various weighting options available. Source code for the calculations is available at https://github.com/mqcomplab/MultipleComparisons.

DeepARV: ensemble deep learning to predict drug-drug interaction of clinical relevance with antiretroviral therapy

Drug-drug interaction (DDI) may result in clinical toxicity or treatment failure of antiretroviral therapy (ARV) or comedications. Despite the high number of possible drug combinations, only a limited number of clinical DDI studies are conducted. Computational prediction of DDIs could provide key evidence for the rational management of complex therapies. Our study aimed to assess the potential of deep learning approaches to predict DDIs of clinical relevance between ARVs and comedications. DDI severity grading between 30,142 drug pairs was extracted from the Liverpool HIV Drug Interaction database. Two feature construction techniques were employed: 1) drug similarity profiles by comparing Morgan fingerprints, and 2) embeddings from SMILES of each drug via ChemBERTa, a transformer-based model. We developed DeepARV-Sim and DeepARV-ChemBERTa to predict four categories of DDI: i) Red: drugs should not be co-administered, ii) Amber: interaction of potential clinical relevance manageable by monitoring/dose adjustment, iii) Yellow: interaction of weak relevance and iv) Green: no expected interaction. The imbalance in the distribution of DDI severity grades was addressed by undersampling and applying ensemble learning. DeepARV-Sim and DeepARV-ChemBERTa predicted clinically relevant DDI between ARVs and comedications with a weighted mean balanced accuracy of 0.729 ± 0.012 and 0.776 ± 0.011, respectively. DeepARV-Sim and DeepARV-ChemBERTa have the potential to leverage molecular structures associated with DDI risks and reduce DDI class imbalance, effectively increasing the predictive ability on clinically relevant DDIs. This approach could be developed for identifying high-risk pairing of drugs, enhancing the screening process, and targeting DDIs to study in clinical drug development.

Reactivities of acrylamide warheads toward cysteine targets: a QM/ML approach to covalent inhibitor design.

Covalent inhibition offers many advantages over non-covalent inhibition, but covalent warhead reactivity must be carefully balanced to maintain potency while avoiding unwanted side effects. While warhead reactivities are commonly measured with assays, a computational model to predict warhead reactivities could be useful for several aspects of the covalent inhibitor design process. Studies have shown correlations between covalent warhead reactivities and quantum mechanic (QM) properties that describe important aspects of the covalent reaction mechanism. However, the models from these studies are often linear regression equations and can have limitations associated with their usage. Applications of machine learning (ML) models to predict covalent warhead reactivities with QM descriptors are not extensively seen in the literature. This study uses QM descriptors, calculated at different levels of theory, to train ML models to predict reactivities of covalent acrylamide warheads. The QM/ML models are compared with linear regression models built upon the same QM descriptors and with ML models trained on structure-based features like Morgan fingerprints and RDKit descriptors. Experiments show that the QM/ML models outperform the linear regression models and the structure-based ML models, and literature test sets demonstrate the power of the QM/ML models to predict reactivities of unseen acrylamide warhead scaffolds. Ultimately, these QM/ML models are effective, computationally feasible tools that can expedite the design of new covalent inhibitors.

Harnessing machine learning to predict cytochrome P450 inhibition through molecular properties.

Cytochrome P450 enzymes are a superfamily of enzymes responsible for the metabolism of a variety of medicines and xenobiotics. Among the Cytochrome P450 family, five isozymes that include 1A2, 2C9, 2C19, 2D6, and 3A4 are most important for the metabolism of xenobiotics. Inhibition of any of these five CYP isozymes causes drug-drug interactions with high pharmacological and toxicological effects. So, the inhibition or non-inhibition prediction of these isozymes is of great importance. Many techniques based on machine learning and deep learning algorithms are currently being used to predict whether these isozymes will be inhibited or not. In this study, three different molecular or substructural properties that include Morgan, MACCS and Morgan (combined) and RDKit of the various molecules are used to train a distinct SVM model against each isozyme (1A2, 2C9, 2C19, 2D6, and 3A4). On the independent dataset, Morgan fingerprints provided the best results, while MACCS and Morgan (combined) achieved comparable results in terms of balanced accuracy (BA), sensitivity (Sn), and Mathews correlation coefficient (MCC). For the Morgan fingerprints, balanced accuracies (BA), Mathews correlation coefficients (MCC), and sensitivities (Sn) against each CYPs isozyme, 1A2, 2C9, 2C19, 2D6, and 3A4 on an independent dataset ranged between 0.81 and 0.85, 0.61 and 0.70, 0.72 and 0.83, respectively. Similarly, on the independent dataset, MACCS and Morgan (combined) fingerprints achieved competitive results in terms of balanced accuracies (BA), Mathews correlation coefficients (MCC), and sensitivities (Sn) against each CYPs isozyme, 1A2, 2C9, 2C19, 2D6, and 3A4, which ranged between 0.79 and 0.85, 0.59 and 0.69, 0.69 and 0.82, respectively.

Abstract 4937: Potential therapeutic targets with anti-inflammatory activity spectrum(COX-2/PGE2) of compounds associated with anti-cancer activityagainst oral and colorectal cancers

Abstract Introduction: COX-2 and PGE2 are prevalent inflammatory cytokines that exert influence on several cancer-related pathways throughout the gastrointestinal tract, spanning from oral cancers to colorectal cancers. One of the tactics being developed for the treatment of oral and colorectal malignancies is the identification of drugs that can selectively target receptors associated with COX-2 and PGE2. Materials and Methods: The objective was to identify potential therapeutic targets for oral cancer and colorectal cancer, specifically focusing on drugs that act as inhibitors or antagonists of the COX2/PGE2 pathway and its related receptor. A method for systematic data mining was created with a total of 32 targets associated with the COX-2/PGE2 pathway which were identified using the KEGG database. These targets were shown to be linked to both the oral cancer pathway and the colorectal cancer pathway. The compounds of interest were obtained from the Chembl25 database and their similarity was assessed using the Tanimoto similarity index, which was generated based on the Morgan fingerprints. The study employed molecular docking techniques to examine the interaction between the chosen drugs and the anticipated targets. A systematic data mining approach and computational methodology were devised utilizing the KNIME platform. Results: A total of 16,900 compounds were extracted for a set of 32 targets. Following the implementation of rule-based filters, the number of compounds was reduced to 13,800. The compounds that have a similarity index greater than 0.7 were chosen for inclusion in final analysis. Among various targets examined, empirical evidence was found to support the inhibition of expected target for PI-3 kinase-mTOR, GSK-3CD1, GSK3-AKT and PI3kinase-AKT pair. However, no evidence was found to support the inhibition of GSK3-Ikappa B-kinase. In order to examine the likelihood of the compound's interaction with anticipated target, molecular docking was employed to explore the interaction of one inhibitor compound for each pair. This investigation was conducted in contrast to the existing inhibitor compound and the subsequent findings were analyzed and discussed. Conclusion: The implemented data mining technique has demonstrated a notable decrease in the duration needed for comparable investigations, enabling the execution of experiments including numerous targets and a substantial number of ligands within a cumulative timeframe of 30 minutes. The findings of this study have the potential to be utilized in the identification of targets and the discovery of lead compounds for the purpose of managing oral as well as colorectal malignancies. Citation Format: Saravanan Sampoornam Pape. Potential therapeutic targets with anti-inflammatory activity spectrum(COX-2/PGE2) of compounds associated with anti-cancer activityagainst oral and colorectal cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4937.

Systematic Approaches for the Encoding of Chemical Groups: A Case Study.

Regulatory authorities aim to organize substances into groups to facilitate prioritization within hazard and risk assessment processes. Often, such chemical groupings are not explicitly defined by structural rules or physicochemical property information. This is largely due to how these groupings are developed, namely, a manual expert curation process, which in turn makes updating and refining groupings, as new substances are evaluated, a practical challenge. Herein, machine learning methods were leveraged to build models that could preliminarily assign substances to predefined groups. A set of 86 groupings containing 2,184 substances as published on the European Chemicals Agency (ECHA) website were mapped to the U.S. Environmental Protection Agency (EPA) Distributed Toxicity Structure Database (DSSTox) content to extract chemical and structural information. Substances were represented using Morgan fingerprints, and two machine learning approaches were used to classify test substances into 56 groups containing at least 10 substances with a structural representation in the data set: k-nearest neighbor (kNN) and random forest (RF), that led to mean 5-fold cross-validation test accuracies (average F1 scores) of 0.781 and 0.853, respectively. With a 9% improvement, the RF classifier was significantly more accurate than KNN (p-value = 0.001). The approach offers promise as a means of the initial profiling of new substances into predefined groups to facilitate prioritization efforts and streamline the assessment of new substances when earlier groupings are available. The algorithm to fit and use these models has been made available in the accompanying repository, thereby enabling both use of the produced models and refitting of these models, as new groupings become available by regulatory authorities or industry.

Construction of An Oral Bioavailability Prediction Model Based on Machine Learning for Evaluating Molecular Modifications

ObjectiveThis study aims to explore the impact of ADME on the Oral Bioavailability (OB) of drugs and to construct a machine learning model for OB prediction. The model is then applied to predict the OB of modified berberine and atenolol molecules to obtain structures with higher OB. MethodsInitially, a drug OB database was established, and corresponding ADME characteristics were obtained. The relationship between ADME and OB was analyzed using machine learning, with Morgan fingerprints serving as molecular descriptors. Compounds from the database were input into Random Forest, XGBoost, CatBoost, and LightGBM machine learning models to train the OB 7prediction model and evaluate its performance. Subsequently, berberine and atenolol were modified using Chemdraw software with ten different substituents for mono-substitution, and chlorine atoms for a full range of double substitutions. The modified molecular structures were converted into the same format as the training set for OB prediction. The predicted OB values of the modified structures of berberine and atenolol were compared. ResultsAn OB database of 386 drugs was obtained. It was found that smaller molecular weight and a higher number of rotatable bonds (ten or less) could potentially lead to higher OB. The four machine learning models were evaluated using MSE, R2 score, MAE, and MFE as metrics, with Random Forest performing the best. The models' predictions for the test set were particularly accurate when OB ranged from 30% to 90%. After mono-substitution and double substitution of berberine and atenolol, the OB of both drugs was significantly improved. ConclusionsThis study found that some ADME properties of molecules do not have an absolute impact on OB. The database played a decisive role in the process of the machine learning OB prediction model, and the performance of the model was evaluated based on predictions within a range of strong generalization ability. In most cases, mono-substitution and double substitution were beneficial for enhancing the OB of berberine and atenolol. In summary, this study successfully constructed a machine learning regression prediction model that can accurately predict drug OB, which can guide drug design to achieve higher OB to some extent.

Machine learning for layer-by-layer nanofiltration membrane performance prediction and polymer candidate exploration

In this study, machine learning-based models were established for layer-by-layer (LBL) nanofiltration (NF) membrane performance prediction and polymer candidate exploration. Four different models, i.e., linear, random forest (RF), boosted tree (BT), and eXtreme Gradient Boosting (XGBoost), were formed, and membrane performance prediction was determined in terms of membrane permeability and selectivity. The XGBoost exhibited optimal prediction accuracy for membrane permeability (coefficient of determination (R2): 0.99) and membrane selectivity (R2: 0.80). The Shapley Additive exPlanation (SHAP) method was utilized to evaluate the effects of different LBL NF membrane fabrication conditions on membrane performances. The SHAP method was also used to identify the relationships between polymer structure and membrane performance. Polymers were represented by Morgan fingerprint, which is an effective description approach for developing modeling. Based on the SHAP value results, two reference Morgan fingerprints were constructed containing atomic groups with positive contributions to membrane permeability and selectivity. According to the reference Morgan fingerprint, 204 potential polymers were explored from the largest polymer database (PoLyInfo). By calculating the similarities between each potential polymer and both reference Morgan fingerprints, 23 polymer candidates were selected and could be further used for LBL NF membrane fabrication with the potential for providing good membrane performance. Overall, this work provided new ways both for LBL NF membrane performance prediction and high-performance polymer candidate exploration. The source code for the models and algorithms used in this study is publicly available to facilitate replication and further research. https://github.com/wangliwfsd/LLNMPP/

Attention-based solubility prediction of polysulfide and electrolyte analysis for lithium–sulfur batteries

During the continuous charge and discharge process in lithium-sulfur batteries, one of the next-generation batteries, polysulfides are generated in the battery’s electrolyte, and impact its performance in terms of power and capacity by involving the process. The amount of polysulfides in the electrolyte could be estimated by the change of the Gibbs free energy of the electrolyte, Delta _{mix}textrm{G} in the presence of polysulfide. However, obtaining Delta _{mix}textrm{G} of the diverse mixtures of components in the electrolyte is a complex and expensive task that shows itself as a bottleneck in optimization of electrolytes. In this work, we present a machine-learning approach for predicting Delta _{mix}textrm{G} of electrolytes. The proposed architecture utilizes (1) an attention-based model (Attentive FP), a contrastive learning model (MolCLR) or morgan fingerprints to represent chemical components, and (2) transformers to account for the interactions between chemicals in the electrolyte. This architecture was not only capable of predicting electrolyte properties, including those of chemicals not used during training, but also providing insights into chemical interactions within electrolytes. It revealed that interactions with other chemicals relate to the logP and molecular weight of the chemicals.

Predicting absolute aqueous solubility by applying a machine learning model for an artificially liquid-state as proxy for the solid-state.

In this study, we use machine learning algorithms with QM-derived COSMO-RS descriptors, along with Morgan fingerprints, to predict the absolute solubility of drug-like compounds. The QM-derived descriptors account for the molecular properties of the solute, i.e., the solute-solute interactions in an artificial-liquid-state (super-cooled liquid), and the solute-solvent interactions in solution. We employ two main approaches to predict solubility: (i) a hypothetical pathway that involves melting the solute at room temperature T = T¯ ([Formula: see text]) and mixing the artificially liquid solute into the solvent ([Formula: see text]). In this approach [Formula: see text] is predicted using machine learning models, and the [Formula: see text] is obtained from COSMO-RS calculations; (ii) direct solubility prediction using machine learning algorithms. The models were trained on a large number of Bayer in-house compounds for which water solubility data is available at physiological pH of 6.5 and ambient temperature. We also evaluated our models using external datasets from a solubility challenge. Our models present great improvements compared to the absolute solubility prediction with the QSAR model for the artificial liquid state as implemented in the COSMOtherm software, for both in-house and external datasets. We are furthermore able to demonstrate the superiority of QM-derived descriptors compared to cheminformatics descriptors. We finally present low-cost alternative models using fragment-based COSMOquick calculations with only marginal reduction in the quality of predicted solubility.