Simplified Molecular Input Line Entry System Research Articles

Diffusion-based generative drug-like molecular editing with chemical natural language

Recently, diffusion models have emerged as a promising paradigm for molecular design and optimization. However, most diffusion-based molecular generative models focus on modeling 2D graphs or 3D geometries, with limited research on molecular sequence diffusion models. The International Union of Pure and Applied Chemistry (IUPAC) names are more akin to chemical natural language than the Simplified Molecular Input Line Entry System (SMILES) for organic compounds. In this work, we apply an IUPAC-guided conditional diffusion model to facilitate molecular editing from chemical natural language to chemical language (SMILES) and explore whether the pre-trained generative performance of diffusion models can be transferred to chemical natural language. We propose DiffIUPAC, a controllable molecular editing diffusion model that converts IUPAC names to SMILES strings. Evaluation results demonstrate that our model outperforms existing methods and successfully captures the semantic rules of both chemical languages. Chemical space and scaffold analysis show that the model can generate similar compounds with diverse scaffolds within the specified constraints. Additionally, to illustrate the model's applicability in drug design, we conducted case studies in functional group editing, analogue design and linker design.

Development and validation of a machine learning model for predicting drug-drug interactions with oral diabetes medications

Diabetes management is often complicated by comorbidities, requiring complex medication regimens that increase the risk of drug-drug interactions (DDIs), potentially compromising treatment outcomes or causing toxicity. Although machine learning (ML) models have made strides in DDI prediction, existing approaches lack specificity for oral diabetes medications and face challenges in interpretability. To address these limitations, we propose a novel ML-based framework utilizing the Simplified Molecular Input Line Entry System (SMILES) to encode structural information of oral diabetes drugs. Using this representation, we developed an XGBoost model, selecting molecular features through LASSO. Our dataset, sourced from DrugBank, included 42 oral diabetes drugs and 1,884 interacting drugs, divided into training, validation, and testing sets. The model identified 606 optimal features, achieving an F1-score of 0.8182. SHAP analysis was employed for feature interpretation, enhancing model transparency and clinical relevance. By predicting adverse DDIs, our model offers a valuable tool for clinical decision-making, aiding safer prescription practices. The 606 critical features provide insights into atomic-level interactions, linking computational predictions with biological experiments. We present a classification model specifically designed for predicting DDIs associated with oral diabetes medications, with an openly accessible web application to support diabetes management in multi-drug regimens and comorbidity settings.

A QSPR Model for Henry's Law Constants of Organic Compounds in Water and Ethanol for Distilled Spirits.

Henry's law describes the vapor-liquid equilibrium for dilute gases dissolved in a liquid solvent phase. Descriptions of vapor-liquid equilibrium allow the design of improved separations in the food and beverage industry. The consumer experience of taste and odor are greatly affected by the liquid and vapor phase behavior of organic compounds. This study presents a machine learning (ML) based model that allows quick, accurate predictions of Henry's law constants (kH) for many common organic compounds. Users input only a Simplified Molecular-Input Line-Entry System (SMILES) string or a common English name, and the model returns Henry's law estimates for compounds in water and ethanol. Training was performed on 5,690 compounds. Training data were gathered from an existing database and were supplemented with quantum mechanical (QM) calculations. An extra trees regression model was generated that predicts kH with a mean absolute error of 1.3 in log space and an R2 of 0.98. The model is applied to common flavor and odor compounds in bourbon whiskey as a test case for food and beverage applications.

In silico aquatic toxicity prediction of chemicals toward Daphnia magna and fathead minnow using Monte Carlo approaches

The fast-increasing use of chemicals led to large numbers of chemical compounds entering the aquatic environment, raising concerns about their potential effects on ecosystems. Therefore, assessment of the ecotoxicological features of organic compounds on aquatic organisms is very important. Daphnia magna and Fathead minnow are two aquatic species that are commonly tested as standard test organisms for aquatic risk assessment and are typically chosen as the biological model for the ecotoxicology investigations of chemical pollutants. Herein, global quantitative structure–toxicity relationship (QSTR) models have been developed to predict the toxicity (pEC(LC)50) of a large dataset comprising 2106 chemicals toward Daphnia magna and Fathead minnow. The optimal descriptor of correlation weights (DCWs) is calculated using the notation of simplified molecular input line entry system (SMILES) and is used to construct QSTR models. Three target functions, TF1, TF2, and TF3 are utilized to generate 12 QSTR models from four splits, and their statistical characteristics are also compared. The designed QSTR models are validated using both internal and external validation criteria and are found to be reliable, robust, and excellently predictive. Among the models, those generated using the TF3 demonstrate the best statistical quality with R 2 values ranging from 0.9467 to 0.9607, Q 2 values ranging from 0.9462 to 0.9603 and RMSE values ranging from 0.3764 to 0.4413 for the validation set. The applicability domain and the mechanistic interpretations of generated models were also discussed.

A new spectral-based index for graphs and its application to polyaromatic compounds

In this study, we introduce a novel index called the modified inverse sum indeg (MISI) energy as an extension of the traditional inverse sum indeg (ISI) energy and neighbor degree sum energy. We devised an algorithm that employs the simplified molecular input line entry system (SMILES) formula of compounds to compute the MISI energy of polycyclic aromatic hydrocarbon (PAH). Additionally, we established the bounds of the MISI energy for certain classes of graphs with fixed number of vertices. A strong correlation between the MISI energy and the total [Formula: see text]-electron energy of PAHs is observed through regression analysis. Remarkably, this correlation outperforms that achieved by the classical ISI energy. These findings highlight the enhanced effectiveness of the MISI energy as a descriptor for capturing significant molecular properties. Moreover, our study establishes a foundation for further theoretical extensions and practical applications in the field of chemical graph theory.

A multi-view feature representation for predicting drugs combination synergy based on ensemble and multi-task attention models

This paper proposes a novel multi-view ensemble predictor model that is designed to address the challenge of determining synergistic drug combinations by predicting both the synergy score value values and synergy class label of drug combinations with cancer cell lines. The proposed methodology involves representing drug features through four distinct views: Simplified Molecular-Input Line-Entry System (SMILES) features, molecular graph features, fingerprint features, and drug-target features. On the other hand, cell line features are captured through four views: gene expression features, copy number features, mutation features, and proteomics features. To prevent overfitting of the model, two techniques are employed. First, each view feature of a drug is paired with each corresponding cell line view and input into a multi-task attention deep learning model. This multi-task model is trained to simultaneously predict both the synergy score value and synergy class label. This process results in sixteen input view features being fed into the multi-task model, producing sixteen prediction values. Subsequently, these prediction values are utilized as inputs for an ensemble model, which outputs the final prediction value. The ‘MVME’ model is assessed using the O’Neil dataset, which includes 38 distinct drugs combined across 39 distinct cancer cell lines to output 22,737 drug combination pairs. For the synergy score value, the proposed model scores a mean square error (MSE) of 206.57, a root mean square error (RMSE) of 14.30, and a Pearson score of 0.76. For the synergy class label, the model scores 0.90 for accuracy, 0.96 for precision, 0.57 for kappa, 0.96 for the area under the ROC curve (ROC-AUC), and 0.88 for the area under the precision-recall curve (PR-AUC).

Cheminformatic Identification of Tyrosyl-DNA Phosphodiesterase 1 (Tdp1) Inhibitors: A Comparative Study of SMILES-Based Supervised Machine Learning Models.

Tyrosyl-DNA phosphodiesterase 1 (Tdp1) repairs damages in DNA induced by abortive topoisomerase 1 activity; however, maintenance of genetic integrity may sustain cellular division of neoplastic cells. It follows that Tdp1-targeting chemical inhibitors could synergize well with existing chemotherapy drugs to deny cancer growth; therefore, identification of Tdp1 inhibitors may advance precision medicine in oncology. Current computational research efforts focus primarily on molecular docking simulations, though datasets involving three-dimensional molecular structures are often hard to curate and computationally expensive to store and process. We propose the use of simplified molecular input line entry system (SMILES) chemical representations to train supervised machine learning (ML) models, aiming to predict potential Tdp1 inhibitors. An open-sourced consensus dataset containing the inhibitory activity of numerous chemicals against Tdp1 was obtained from Kaggle. Various ML algorithms were trained, ranging from simple algorithms to ensemble methods and deep neural networks. For algorithms requiring numerical data, SMILES were converted to chemical descriptors using RDKit, an open-sourced Python cheminformatics library. Out of 13 optimized ML models with rigorously tuned hyperparameters, the random forest model gave the best results, yielding a receiver operating characteristics-area under curve of 0.7421, testing accuracy of 0.6815, sensitivity of 0.6444, specificity of 0.7156, precision of 0.6753, and F1 score of 0.6595. Ensemble methods, especially the bootstrap aggregation mechanism adopted by random forest, outperformed other ML algorithms in classifying Tdp1 inhibitors from non-inhibitors using SMILES. The discovery of Tdp1 inhibitors could unlock more treatment regimens for cancer patients, allowing for therapies tailored to the patient's condition.

Drug-Target Binding Affinity Prediction in a Continuous Latent Space Using Variational Autoencoders.

Accurate prediction of Drug-Target binding Affinity (DTA) is a daunting yet pivotal task in the sphere of drug discovery. Over the years, a plethora of deep learning-based DTA models have emerged, rendering promising results in predicting the binding affinities between drugs and their target proteins. However, in contrast to the conventional approach of modeling binding affinity in vector spaces, we propose a more nuanced modeling process in a continuous space to account for the diversity of input samples. Initially, the drug is encoded using the Simplified Molecular Input Line Entry System (SMILES), while the target sequences are characterized via a pretrained language model. Subsequently, highly correlative information is extracted utilizing residual gated convolutional neural networks. In a departure from existing deep learning-based models, our model learns the hidden representations of the drugs and targets jointly. Instead of employing two vectors, our hidden representations consist of two Gaussian distributions. To validate the effectiveness of our proposal, we conducted evaluations on commonly utilized benchmark datasets. The experimental outcomes corroborated that our method surpasses the state-of-the-art vectorial representation methods in terms of performance. This approach, therefore, offers potential enhancements in the precision of DTA predictions, potentially contributing to more efficient drug discovery processes.

MGFmiRNAloc: Predicting miRNA Subcellular Localization Using Molecular Graph Feature and Convolutional Block Attention Module.

MiRNA has distinct physiological functions at various cellular locations. However, few effective computational methods for predicting the subcellular location of miRNA exist, thereby leaving considerable room for improvement. Accordingly, our study proposes the MGFmiRNAloc simplified molecular input line entry system (SMILES) format as a new approach for predicting the subcellular localization of miRNA. Additionally, the graphical convolutional network (GCN) technique was employed to extract the atomic nodes and topological structure of a single base, thereby constructing RNA sequence molecular map features. Subsequently, the channel attention and spatial attention mechanisms (CBAM) were designed to mine deeper for more efficient information. Finally, the prediction module was used to detect the subcellular localization of miRNA. The 10-fold cross-validation and independent test set experiments demonstrate that MGFmiRNAloc outperforms the most sophisticated methods. The results indicate that the new atomic level feature representation proposed in this study could overcome the limitations of small samples and short miRNA sequences, accurately predict the subcellular localization of miRNAs, and be extended to the subcellular localization of other sequences.

Prediction of Glass Transition Temperature of Polymers Using Simple Machine Learning.

Polymer materials have garnered significant attention due to their exceptional mechanical properties and diverse industrial applications. Understanding the glass transition temperature (Tg) of polymers is critical to prevent operational failures at specific temperatures. Traditional methods for measuring Tg, such as differential scanning calorimetry (DSC) and dynamic mechanical analysis, while accurate, are often time-consuming, costly, and susceptible to inaccuracies due to random and uncertain factors. To address these limitations, the aim of the present study is to investigate the potential of Simplified Molecular Input Line Entry System (SMILES) as descriptors in simple machine learning models to predict Tg efficiently and reliably. Five models were utilized: k-nearest neighbors (KNNs), support vector regression (SVR), extreme gradient boosting (XGBoost), artificial neural network (ANN), and recurrent neural network (RNN). SMILES descriptors were converted into numerical data using either One Hot Encoding (OHE) or Natural Language Processing (NLP). The study found that SMILES inputs with fewer than 200 characters were inadequate for accurately describing compound structures, while inputs exceeding 200 characters diminished model performance due to the curse of dimensionality. The ANN model achieved the highest R2 value of 0.79; however, the XGB model, with an R2 value of 0.774, exhibited the highest stability and shorter training times compared to other models, making it the preferred choice for Tg prediction. The efficiency of the OHE method over NLP was demonstrated by faster training times across the KNN, SVR, XGB, and ANN models. Validation of new polymer data showed the XGB model's robustness, with an average prediction deviation of 9.76 from actual Tg values. These findings underscore the importance of optimizing SMILES conversion methods and model parameters to enhance prediction reliability. Future research should focus on improving model accuracy and generalizability by incorporating additional features and advanced techniques. This study contributes to the development of efficient and reliable predictive models for polymer properties, facilitating the design and application of new polymer materials.

Unlocking comprehensive molecular design across all scenarios with large language model and unordered chemical language.

Molecular generation stands at the forefront of AI-driven technologies, playing a crucial role in accelerating the development of small molecule drugs. The intricate nature of practical drug discovery necessitates the development of a versatile molecular generation framework that can tackle diverse drug design challenges. However, existing methodologies often struggle to encompass all aspects of small molecule drug design, particularly those rooted in language models, especially in tasks like linker design, due to the autoregressive nature of large language model-based approaches. To empower a language model for a wider range of molecular design tasks, we introduce an unordered simplified molecular-input line-entry system based on fragments (FU-SMILES). Building upon this foundation, we propose FragGPT, a universal fragment-based molecular generation model. Initially pretrained on extensive molecular datasets, FragGPT utilizes FU-SMILES to facilitate efficient generation across various practical applications, such as de novo molecule design, linker design, R-group exploration, scaffold hopping, and side chain optimization. Furthermore, we integrate conditional generation and reinforcement learning (RL) methodologies to ensure that the generated molecules possess multiple desired biological and physicochemical properties. Experimental results across diverse scenarios validate FragGPT's superiority in generating molecules with enhanced properties and novel structures, outperforming existing state-of-the-art models. Moreover, its robust drug design capability is further corroborated through real-world drug design cases.

Molecular interaction analysis of ferulic acid (4-hydroxy-3-methoxycinnamic acid) as main bioactive compound from palm oil waste against MCF-7 receptors: An in silico study.

Ferulic acid (4-hydroxy-3-methoxycinnamic acid) is a phytochemical compound that is commonly found in conjugated forms within mono-, di-, polysaccharides and other organic compounds in cell walls of grain, fruits, and vegetables. This compound is highly abundant in the palm oil waste. The aim of the study was to predict the anticancer activity of ferulic acid against the breast cancer cell lines (MCF-7) receptors through a computational analysis. MCF-7 receptors with PDB IDs of 1R5K, 2IOG, 4IV2, 4IW6, 5DUE, 5T92, and 5U2B were selected based on the Simplified Molecular Input Line Entry System (SMILES) similarity of the native ligand. Thereafter, the protein was prepared on Chimera 1.16 and docked with ferulic acid on Autodock Vina 1.2.5. The ligand-protein complex interaction was validated by computing the root mean square fluctuation (RMSF) and radius of gyration (Rg) through molecular dynamic simulation. In addition, an absorption, distribution, metabolism, excretion, and toxicity (ADMET) prediction was performed on ferulic acid using the pkCSM platform. The molecular docking revealed that the ferulic acid could interact with all receptors as indicated by the affinity energy <-5 kcal/mol. The compound had the most optimum interaction with receptor 2IOG (affinity energy=-6.96 kcal/mol), involving hydrophobic interaction (n=12) and polar hydrogen interaction (n=4). The molecular dynamic simulation revealed that the complex had an RMSF of 1.713 Å with a fluctuation of Rg value around 1.000 Å. The ADMET properties of ferulic acid suggested that the compound is an ideal drug candidate. In conclusion, this study suggested that ferulic acid, which can be isolated from palm oil waste, has the potential to interact with MCF-7 receptors.

The Block Copolymer Phase Behavior Database.

The Block Copolymer Database (BCDB) is a platform that allows users to search, submit, visualize, benchmark, and download experimental phase measurements and their associated characterization information for di- and multiblock copolymers. To the best of our knowledge, there is no widely accepted data model for publishing experimental and simulation data on block copolymer self-assembly. This proposed data schema with traceable information can accommodate any number of blocks and at the time of publication contains over 5400 block copolymer total melt phase measurements mined from the literature and manually curated and simulation data points of the phase diagram generated from self-consistent field theory that can rapidly be augmented. This database can be accessed via the Community Resource for Innovation in Polymer Technology (CRIPT) web application and the Materials Data Facility. The chemical structure of the polymer is encoded in BigSMILES, an extension of the Simplified Molecular-Input Line-Entry System (SMILES) into the macromolecular domain, and the user can search repeat units and functional groups using the SMARTS search syntax (SMILES Arbitrary Target Specification). The user can also query characterization and phase information using Structured Query Language (SQL) and download custom sets of block copolymer data to train machine learning models. Finally, a protocol is presented in which GPT-4, an AI-powered large language model, can be used to rapidly screen and identify block copolymer papers from the literature using only the abstract text and determine whether they have BCDB data, allowing the database to grow as the number of published papers on the World Wide Web increases. The F1 score for this model is 0.74. This platform is an important step in making polymer data more accessible to the broader community.

Comparison of Machine Learning Approaches for Prediction of the Equivalent Alkane Carbon Number for Microemulsions Based on Molecular Properties.

The chemical properties of oils are vital in the design of microemulsion systems. The hydrophilic-lipophilic difference equation used to predict microemulsions' phase behavior expresses the oils' physiochemical properties as the equivalent alkane carbon number (EACN). The experimental determination of EACN requires knowledge of the temperature dependence of the microemulsion system and the effects of different surfactant concentrations. Thus, the experimental determination is time-intensive and tedious, requiring days to months for proper separations. Furthermore, the experiments require high purity of chemicals because microemulsions are sensitive to impurities. Our work focuses on the quick and reliable predictions of the EACN with machine learning (ML) models. Due to the immaturity of ML chemical predictions, we compare three graph neural networks (GNNs) and a gradient-boosted tree algorithm, known as XGBoost. The GNNs use the molecular structures represented as simplified molecular-input line-entry system (SMILES) codes for the initial input, which allows us to assess whether geometry optimization is necessary for reliable results. The XGBoost model also begins with the SMILES representations of the molecules but uses molecular descriptors instead of geometry optimizations. The best model tested (crystal graph convolutional neural network with Merck molecular force field-94) has an error of 1.15 EACN units of the true EACN for unknown data with the errors skewed toward zero and an R2 score of 0.9.

Toxicity Prediction for Immune Thrombocytopenia Caused by Drugs Based on Logistic Regression with Feature Importance

Background: One of the problems in drug discovery that can be solved by artificial intelligence is toxicity prediction. In drug-induced immune thrombocytopenia, toxicity can arise in patients after five to ten days by significant bleeding caused by drugdependent antibodies. In clinical trials, when this condition occurs, all the drugs consumed by patients should be stopped, although sometimes this is not possible, especially for older patients who are dependent on their medication. Therefore, being able to predict toxicity in drug-induced immune thrombocytopenia is very important. Computational technologies, such as machine learning, can help predict toxicity better than empirical techniques owing to the lower cost and faster processing. Objective: Previous studies used the KNN method. However, the performance of these approaches needs to be enhanced. This study proposes a Logistic Regression to improve accuracy scores. Methods: In this study, we present a new model for drug-induced immune thrombocytopenia using a machine learning method. Our model extracts several features from the Simplified Molecular Input Line Entry System (SMILES). These features were fused and cleaned, and the important features were selected using the SelectKBest method. The model uses a Logistic Regression that is optimized and tuned by the Grid Search Cross Validation. Results: The highest accuracy occurred when using features from PADEL, CDK, RDKIT, MORDRED, BLUEDESC combinations, resulting in an accuracy of 80%. Conclusion: Our proposed model outperforms previous studies in accuracy categories. The information and source code is accessible online at Github: https://github.com/Osphanie/Thrombocytopenia

Multicellular target QSAR models for predicting of novel inhibitors against pancreatic cancer by Monte Carlo approach

Pancreatic cancer is one of the most fatal malignant tumors, with a poor prognosis. Nowadays, computational methods for designing and predicting anticancer agents have gained increased attention. In this work, a QSAR study is conducted to further investigate the activity of 5797 agents against pancreatic cancer cell lines. QSAR models are a system of classification approaches that are used to construct and validate the predictive potential of created models. This approach demonstrates the prediction in the active (1) and inactive (−1) format. Here, the classification QSAR technique is employed to construct and evaluate models. The classification QSAR model has been constructed by the Monte Carlo method based on the representation of the molecular structure by SMILES (simplified molecular input line entry systems) using the CORAL software. Four random splits of data by making four sets such as active training, passive training, calibration, and validation sets are assessed. The numerical values of sensitivity, specificity, and accuracy for the validation set of splits 1 to 4 are in the range of 0.7726–0.7901, 0.9007–0.9220, and 0.8548–0.8730 respectively. The Matthews correlation coefficient (MCC) of these models for external validation sets is found to be in the range of 0.6832 to 0.7215.

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.

Advancements in hand-drawn chemical structure recognition through an enhanced DECIMER architecture

Accurate recognition of hand-drawn chemical structures is crucial for digitising hand-written chemical information in traditional laboratory notebooks or facilitating stylus-based structure entry on tablets or smartphones. However, the inherent variability in hand-drawn structures poses challenges for existing Optical Chemical Structure Recognition (OCSR) software. To address this, we present an enhanced Deep lEarning for Chemical ImagE Recognition (DECIMER) architecture that leverages a combination of Convolutional Neural Networks (CNNs) and Transformers to improve the recognition of hand-drawn chemical structures. The model incorporates an EfficientNetV2 CNN encoder that extracts features from hand-drawn images, followed by a Transformer decoder that converts the extracted features into Simplified Molecular Input Line Entry System (SMILES) strings. Our models were trained using synthetic hand-drawn images generated by RanDepict, a tool for depicting chemical structures with different style elements. A benchmark was performed using a real-world dataset of hand-drawn chemical structures to evaluate the model's performance. The results indicate that our improved DECIMER architecture exhibits a significantly enhanced recognition accuracy compared to other approaches.Scientific contributionThe new DECIMER model presented here refines our previous research efforts and is currently the only open-source model tailored specifically for the recognition of hand-drawn chemical structures. The enhanced model performs better in handling variations in handwriting styles, line thicknesses, and background noise, making it suitable for real-world applications. The DECIMER hand-drawn structure recognition model and its source code have been made available as an open-source package under a permissive license.Graphical

Identifying High-Quality Leads among Screened Anticancerous Compounds Using SMILES Representations.

Cancer is a lethal disease that affects numerous people worldwide. Chemotherapy stands as one of the most effective treatment regimens to combat cancer. Nevertheless, anticancer drugs face a high failure rate due to safety and efficacy issues. Drug failure could be subdued by instigating drug leads with reduced toxicity and enhanced efficacy. Computer-aided drug discovery endorses drug leads in manoeuvring protein and ligand structures or representations. Simplified molecular input line entry system (SMILES) is a linear notation representing the three-dimensional structure of a molecule using symbols and alphanumeric characters. SMILES representation hoards rings and scaffold structures in its depiction. Mining ring and scaffold patterns from molecular SMILES would assist in ascertaining biological properties based on molecular patterns. Moreover, the emergence of artificial intelligence (AI) technologies would accelerate identification of efficient anticancer drug leads. AI algorithms proclaimed for their pattern recognition ability could be employed for identifying molecular patterns from SMILES representation, thereby enabling property prediction. Consequently, we developed a multilayer perceptron (MLP) model for the prediction of anticancer activity using SMILES of NCI-60 cancer growth inhibition data. Furthermore, the top 8 frequent scaffolds were identified on preliminary analysis of cancer growth inhibition data and ChEMBL drugs. The developed MLP model classified anticancer and nonanticancer compounds with a classification accuracy of 0.92. Also, benchmarking of the developed model with machine learning algorithms exhibited better performance of the MLP model.

Can large language models understand molecules?

PurposeLarge Language Models (LLMs) like Generative Pre-trained Transformer (GPT) from OpenAI and LLaMA (Large Language Model Meta AI) from Meta AI are increasingly recognized for their potential in the field of cheminformatics, particularly in understanding Simplified Molecular Input Line Entry System (SMILES), a standard method for representing chemical structures. These LLMs also have the ability to decode SMILES strings into vector representations.MethodWe investigate the performance of GPT and LLaMA compared to pre-trained models on SMILES in embedding SMILES strings on downstream tasks, focusing on two key applications: molecular property prediction and drug-drug interaction prediction.ResultsWe find that SMILES embeddings generated using LLaMA outperform those from GPT in both molecular property and DDI prediction tasks. Notably, LLaMA-based SMILES embeddings show results comparable to pre-trained models on SMILES in molecular prediction tasks and outperform the pre-trained models for the DDI prediction tasks.ConclusionThe performance of LLMs in generating SMILES embeddings shows great potential for further investigation of these models for molecular embedding. We hope our study bridges the gap between LLMs and molecular embedding, motivating additional research into the potential of LLMs in the molecular representation field. GitHub: https://github.com/sshaghayeghs/LLaMA-VS-GPT.