Geometry-Based Molecular Generation With Deep Constrained Variational Autoencoder.

We report a novel framework for achieving fragment-based molecular design using pixel convolutional neural network (PixelCNN) combined with the simplified molecular input line entry system (SMILES) as molecular representation. While a widely used recurrent neural network (RNN) assumes monotonically decaying correlations in strings, PixelCNN captures a periodicity among characters of SMILES. Thus, PixelCNN provides us with a novel solution for the analysis of chemical space by extracting the periodicity of molecular structures that will be buried in SMILES. Moreover, this characteristic enables us to generate molecules by combining several simple building blocks, such as a benzene ring and side-chain structures, which contributes to the effective exploration of chemical space by step-by-step searching for molecules from a target fragment. In conclusion, PixelCNN could be a powerful approach focusing on the periodicity of molecules to explore chemical space for the fragment-based molecular design.

Molecular property prediction with deep learning has accelerated drug discovery and retrosynthesis. However, the shortage of labeled molecular data and the challenge of generalizing across the vast chemical spaces pose significant hurdles for leveraging deep learning in molecular property prediction. This study proposes a self-supervised framework designed to acquire a Simplified Molecular Input Line Entry System (SMILES) representation, which we have dubbed Simple SMILES contrastive learning (SimSon). SimSon was pre-trained using unlabeled SMILES data through contrastive learning to grasp the SMILES representations. Our findings demonstrate that contrastive learning with randomized SMILES enriches the ability of the model to generalize and its robustness as it captures the global semantic context at the molecular level. In downstream tasks, SimSon performs competitively when compared to graph-based methods and even outperforms them on certain benchmark datasets. These results indicate that SimSon effectively captures structural information from SMILES, exhibiting remarkable generalization and robustness. The potential applications of SimSon extend to bioinformatics and cheminformatics, encompassing areas such as drug discovery and drug-drug interaction prediction. The source code is available at https://github.com/lee00206/SimSon.

Several recent reports have shown that long short-term memory generative neural networks (LSTM) of the type used for grammar learning efficiently learn to write Simplified Molecular Input Line Entry System (SMILES) of druglike compounds when trained with SMILES from a database of bioactive compounds such as ChEMBL and can later produce focused sets upon transfer learning with compounds of specific bioactivity profiles. Here we trained an LSTM using molecules taken either from ChEMBL, DrugBank, commercially available fragments, or from FDB-17 (a database of fragments up to 17 atoms) and performed transfer learning to a single known drug to obtain new analogs of this drug. We found that this approach readily generates hundreds of relevant and diverse new drug analogs and works best with training sets of around 40,000 compounds as simple as commercial fragments. These data suggest that fragment-based LSTM offer a promising method for new molecule generation.

Deep learning models for the quantitative structure-property relationship (QSPR) have traditionally encountered challenges related to limited interpretability and generalizability. In this study, we present the simplified molecular input line entry system (SMILES) token additivity (STA) model for accurately predicting fuel properties, which takes SMILES as input and employs stacked multihead self-attention encoders to extract molecular structural information. This model provides insights into the structure-property relationships by quantifying the contributions of individual tokens to target properties. Furthermore, since the STA model operates without handcrafted molecular fingerprints, it is capable of generalizing to a broad spectrum of structure-related properties. To validate the model's efficacy, seven critical fuel properties of standard enthalpy of formation (ΔfH°), entropy (S), isobaric heat capacity (Cp), cetane number (CN), boiling point (BP), melting point (MP), and flash point (FP) were tested. The 10-fold cross-validation demonstrated outstanding predictive accuracy, with mean absolute errors of 1.86 kcal/mol (ΔfH°), 0.62 kcal/mol/K (S), and 1.82 kcal/mol/K (Cp), alongside root-mean-square errors (RMSE) of 4.90 (CN), 11.27 °C (BP), 14.09 °C (MP), and 9.47 °C (FP). All properties achieved R2 values exceeding 0.95. The results demonstrate that it achieves predictive accuracy comparable to conventional machine learning models relying on sophisticated feature engineering while also identifying the effect of key tokens on ΔfH° and CN.

Angiotensin Converting Enzyme (ACE) inhibition has been a promising avenue for anti-hypertensive drug development. Our study investigated the inhibitory potential of bioactive compounds derived from six medicinal plants (Allium sativum L., Zingiber officinale Roscoe, Acalypha godseffiana Mast., Moringa oleifera Lam., Vernonia amygdalina Delile, and Rauvolfia vomitoria Afzel.) against ACE using in silico methods. Thirty-one (31) bioactive compounds were screened while Ramipril, and Enalapril were employed as control drugs. 3D structures and canonical Simplified Molecular Input Line Entry System (SMILES) of the bioactive compounds and control drugs were obtained from the PubChem online server. Drug-likeness assessment of the bioactive compounds and protein-ligand docking of successful compounds were conducted using SwissADME online server and AutoDock Vina software. ADMET (absorption, distribution, metabolism, excretion, toxicity) analysis was also done to evaluate the suitability of the hit ligands for further drug development. Of the 31 compounds screened, 17 passed at least four of the five standard rules of drug-likeness determination, while the control drugs (Ramipril and Enalapril) failed one of the rules. Ajmaline, Apigenin, Quercetin, Cryptolepine, Luteolin, Hydroxyvernolide, Kaempferol and Vernodalol had higher binding energies of -9.6 kcal/mol, -8.7 kcal/mol, -8.5 kcal/mol, -8.4 kcal/mol, -8.4 kcal/mol, -8.3 kcal/mol, -8.3 kcal/mol and -7.8 kcal/mol, respectively than Ramipril and Enalapril (-7.6 kcal/mol, and -7.5 kcal/mol). The higher binding energies and the stability of their binding interactions denote these hit ligands as potential antihypertensive drugs targeting ACE. However, wet lab experimental investigation is necessary to validate the inhibitory activity of these compounds and elucidate their mechanisms of action.

Chemical language models (CLMs) can be employed to design molecules with desired properties. CLMs generate new chemical structures in the form of textual representations, such as the simplified molecular input line entry systems (SMILES) strings, in a rule-free manner. However, the quality of these de novo generated molecules is difficult to assess a priori. In this study, we apply the perplexity metric to determine the degree to which the molecules generated by a CLM match the desired design objectives. This model-intrinsic score allows identifying and ranking the most promising molecular designs based on the probabilities learned by the CLM. Using perplexity to compare “greedy” (beam search) with “explorative” (multinomial sampling) methods for SMILES generation, certain advantages of multinomial sampling become apparent. Additionally, perplexity scoring is performed to identify undesired model biases introduced during model training and allows the development of a new ranking system to remove those undesired biases.

This work presents Fragment Graph DataBase (FGDB), a graph database of ligand fragments extracted and generated from the protein entries available in the Protein Data Bank (PDB). FGDB is meant to support and elicit campaigns of fragment-based drug design, by enabling users to query it in order to construct ad hoc, target-specific libraries. In this regard, the database features more than 17 000 fragments, typically small, highly soluble and chemically stable molecules expressed via their canonical Simplified Molecular Input Line Entry System (SMILES) representation. For these fragments, the database provides information related to their contact frequencies with the amino acids, the ligands they are contained in and the proteins the latter bind to. The graph database can be queried via standard web forms and textual searches by a number of identifiers (SMILES, ligand and protein PDB ids) as well as via graphical queries that can be performed against the graph itself, providing users with an intuitive and effective view upon the underlying biological entities. Further search mechanisms via advanced conjunctive/disjunctive/negated textual queries are also possible, in order to allow scientists to look for specific relationships and export their results for further studies. This work also presents two sample use cases where maternal embryonic leucine zipper kinase and mesotrypsin are used as a target, being proteins of high biomedical relevance for the development of cancer therapies.Database URL: http://biochimica3.bio.uniroma3.it/fragments-web/

Accurate prediction of drug-induced adverse drug reactions (ADRs) is crucial for drug safety evaluation, as it directly impacts public health and safety. While various models have shown promising results in predicting ADRs, their accuracy still needs improvement. Additionally, many existing models often lack interpretability when linking molecular structures to specific ADRs and frequently rely on manually selected molecular fingerprints, which can introduce bias. To address these challenges, we propose ToxBERT, an efficient transformer encoder model that leverages attention and masking mechanisms for simplified molecular input line entry system (SMILES) representations. Our results demonstrate that ToxBERT achieved area under the receiver operating characteristic curve (AUROC) scores of 0.839, 0.759, and 0.664 for predicting drug-induced QT prolongation (DIQT), rhabdomyolysis, and liver injury, respectively, outperforming previous studies. Furthermore, ToxBERT can identify drug substructures that are closely associated with specific ADRs. These findings indicate that ToxBERT is not only a valuable tool for understanding the mechanisms underlying specific drug-induced ADRs but also for mitigating potential ADRs in the drug discovery pipeline.

Background: Sickle cell disease is an autosomal recessive disorder caused due to point mutation on a single gene on chromosome 11 resulting in substitution of valine in place of glutamic acid at sixth position of beta-chain of adult hemoglobin (HbA). The purpose of the current study is to evaluate the effectiveness of phytochemicals (Rutin and its derivatives) for antisickling effects. Methods: The structure of Rutin is obtained in Simplified Molecular Input Line Entry System (SMILES) format and is subjected to chemical similarity search using Extended-connectivity fingerprints (ECFP4) fingerprints. The Tanimoto coefficient is used as the similarity metric and top compounds obtained on chemical similarity search are further subjected to molecular docking based screening analysis. A local database of phytochemicals is created using the IMPAAT database. Molecular docking approach is implemented using iGEMDOCK for screening and ranking of Rutin and its derivatives. The docking is implemented with standard docking protocol with population size of 200 and number of generations= 70. Results: All the ligand molecules showed acceptable binding energies with the 2HBS receptor molecule ranging between -124.87 to-79.59kcal/mol. Koprutin showed the highest binding affinity against the 2HBS (2HBS-DEOXYHEMOGLOBIN S) and which can be further developed as an antisickling agent. Conclusion: The integration of insilico modelling techniques with the wet lab based experimental methods can significantly reduce the cost in terms of time and other resources. Further the intricacies of molecular interactions taking place between ligands and receptors can be better understood with the help of molecular modelling techniques.

G-SIMS and SMILES: Simulated fragmentation pathways for identification of complex molecules, amino acids and peptides

Co-evolutions of correlations for QSAR of toxicity of organometallic and inorganic substances: An unexpected good prediction based on a model that seems untrustworthy

The international union of pure and applied chemistry (IUPAC) name nomenclature constitutes a universally recognized standard naming system for allocating names to chemical compounds and is a human-friendly, substructure molecular language. Simplified molecular input line entry system (SMILES) string is currently the most popular molecular representation language and is a computer-friendly, atomic-level molecular language. Considering the readability of IUPAC name and the advantages of SMILES string, it becomes significant to investigate the distinctions of these two molecular languages in term of molecular generation and regression/classification tasks. Thus, we have developed a chemical language model named IUPAC-GPT. Besides molecular generation, we have also incorporated the freezing of IUPAC-GPT model parameters and the attachment of trainable lightweight networks for fine-tuning regression/classification tasks. The results indicate that pre-trained IUPAC-GPT can grasp general knowledge that can be effectively transferred to downstream tasks such as molecular generation, binary classification, and property regression prediction. Furthermore, when utilizing the same configuration, IUPAC-GPT exhibited superior performance compared to the smilesGPT model in term of some property prediction tasks. Overall, transformer-like language models pretrained on IUPAC corpora emerge as promising alternatives, offering improved performance in terms of interpretability and semantic abstraction (chemical groups and modifications) when compared to models pretrained on SMILES corpora.

This paper proposes a multi-task deep learning model for determining drug combination synergistic by simultaneously output synergy scores and synergy class labels. Initially, the two drugs are represented using a Simplified Molecular-Input Line-Entry (SMILE) system. Chemical structural features of the drugs are extracted from the SMILE using the RedKit package. Additionally, an improved Multi-view representation is proposed to extract graph-based drug features. Furthermore, the cancer cell line is represented by gene expression. Then, a three fully connected layers are learned to extract cancer cell line features. To investigate the impact of drug interactions on cell lines, the drug interaction features are extracted from a pretrained drugs interaction network and fed into an attention mechanism along with the cancer cell line features, resulting in the output of affected cancer cell line features. Subsequently, the drug and cell line features are concatenated and fed into an attention mechanism, which produces a two-feature representation for the two predicted tasks. The relationship between the two tasks is learned using the cross-stitch algorithm. Finally, each task feature is inputted into a fully connected subnetwork to predict the synergy score and synergy label. The proposed model ‘MutliSyn’ is evaluated using the O'Neil cancer dataset, comprising 38 unique drugs combined to form 22,737 drug combination pairs, tested on 39 cancer cell lines. For the synergy score, the model achieves a mean square error (MSE) of 219.14, a root mean square error (RMSE) of 14.75, and a Pearson score of 0.76. Regarding the synergy class label, the model achieves an area under the ROC curve (ROC-AUC) of 0.95, an area under the precision-recall curve (PR-AUC) of 0.85, precision of 0.93, kappa of 0.61, and accuracy of 0.90.

Machine learning as a tool for chemical space exploration broadens horizons to work with known and unknown molecules. At its core lies molecular representation, an essential key to improve learning about structure-property relationships. Recently, contrastive frameworks have been showing impressive results for representation learning in diverse domains. Therefore, this paper proposes a contrastive framework that embraces multimodal molecular data. Specifically, our approach jointly trains a graph encoder and an encoder for the simplified molecular-input line-entry system (SMILES) string to perform the contrastive learning objective. Since SMILES is the basis of our method, i.e., we built the molecular graph from the SMILES, we call our framework as SMILES Contrastive Learning (SMICLR). When stacking a nonlinear regressor on the SMICLR's pretrained encoder and fine-tuning the entire model, we reduced the prediction error by, on average, 44% and 25% for the energetic and electronic properties of the QM9 data set, respectively, over the supervised baseline. We further improved our framework's performance when applying data augmentations in each molecular-input representation. Moreover, SMICLR demonstrated competitive representation learning results in an unsupervised setting.

Abstract—Predicting and understanding the potential side effects of pharmaceutical drugs is a formidable challenge in the realm of medical science. The conventional process of assessing drug safety, primarily reliant on manual clinical testing [3] and post-market surveillance [4], [5], is not only arduous but also time-consuming, making it an impediment to rapid drug development and patient well-being. This paper introduces an innovative approach that harnesses the power of advanced machine learning techniques to address this challenge [6], [7], [8], [10], [9]. Our research delves into the intricate web of drug interactions within the human body and the complex factors contributing to adverse reactions. It is a realm where the exact mechanisms that trigger side effects often remain elusive, and the prevalence of rare and severe reactions complicates the task. Our approach encompasses the integration of diverse data sources, including drug characteristics, generic names, molecular structures (SMILES) [11], [12], and drug category [13]. Thereby providing a comprehensive understanding of the complex relationships between these factors and adverse reactions, unifying these disparate pieces of information, we aim to unveil hidden patterns and relationships that can significantly enhance the accuracy of drug side effect predictions. In this study, we employ the Random Forest classification model, known for its robustness and interpretability, to make predictions that not only incorporate the broad spectrum of drug-related factors [14] but also ensure the inclusion of rare and severe side effects in the assessment [15]. To address the challenge of handling non-numerical data, we employ methodologies to convert features such as SMILES structures, Drug generic names, and others into meaningful numerical descriptors, enabling their seamless integration into the predictive model. Our research does not only seek to enhance drug safety evaluation but also endeavors to bridge the chasm between the unpredictability of drug side effects and the need for a more efficient and informed drug development process. With the potential to revolutionize drug safety practices, this research has far-reaching implications in the domain of patient care and drug industry decision-making, promising a safer and more efficient future for the pharmaceutical research industry. Index Terms—SMILES (Simplified Molecular Input Line En- try System), Indication or Drug Target Diseases, Drug Side Effect, Random Forest, decision tree.