Multimodal fusion with relational learning for molecular property prediction

Artificial intelligence (AI) has grown widely into diverse areas of research over the last decade. Publications distributed over the years disclose that biochemistry and analytical chemistry are incorporating AI to the utmost level. Enormous complex data sets are produced by analytical instruments comprising a treasure of information which is very helpful to analytical chemists in characterization. Despite this, we lack capabilities in data analysis and algorithms, which restricts our aptitude to discover and employ this data entirely. Machine learning and AI are being instigated to deal with this current challenge by accelerating several other applications in analytical chemistry. A critical valuation of illustrative reports amalgamating AI with analytical chemistry, spectroscopies, and several sensors has been discussed widely. This comprehensive review evaluates the AI advancement in data interpretation, the assistance of AI in analyte identification and quantification in proteomics and metabolomics, AI databases for primary and secondary metabolites information, regulatory and ethical guidelines in analytical chemistry incorporating AI, machine learning applications in drug discovery, agriculture, cosmetic, and food and separation techniques. In summary, this current review puts forward a thorough examination concerning the advancements of AI in different domains of chemistry and strives to facilitate comprehension of its forthcoming trajectories.

In recent years, machine learning models have shown substantial progress in predicting molecular properties. However, integrating molecular graph structures with sequence information continues to present a significant challenge. In this paper, we introduce Multi-MoleScale, a novel multi-scale framework designed to address this challenge. By combining Graph Contrastive Learning (GCL) with sequence-based models like BERT, Multi-MoleScale enhances the prediction of molecular properties by capturing both structural and contextual representations of molecules. Specifically, the model leverages GCL to effectively capture the intrinsic graph-based features of molecules while utilizing BERT’s pretraining capabilities to learn the contextual relationships within molecular sequences. The contrastive learning component enables Multi-MoleScale to distinguish between relevant and irrelevant molecular features, thereby enhancing its predictive accuracy across diverse molecular types. To assess the performance of our method, we conducted experiments on several widely used public datasets, including 12 molecular property datasets, the ADMET dataset, and 14 breast cancer cell line datasets. The results show that Multi-MoleScale consistently outperforms existing deep learning and self-supervised learning approaches. Notably, the model does not require handcrafted features, making it highly adaptable and versatile for a variety of molecular discovery tasks. This makes Multi-MoleScale a promising tool for applications in drug discovery, materials science, and other molecular research fields. Our data and code are available at https://github.com/pdssunny/Multi-MoleScale. Supplementary InformationThe online version contains supplementary material available at 10.1186/s13321-025-01126-w.

Molecular property prediction has become essential inacceleratingadvancements in drug discovery and materials science. Graph NeuralNetworks have recently demonstrated remarkable success in molecularrepresentation learning; however, their broader adoption is impededby two significant challenges: (1) data scarcity and constrained modelgeneralization due to the expensive and time-consuming task of acquiringlabeled data and (2) inadequate initial node and edge features thatfail to incorporate comprehensive chemical domain knowledge, notablyorbital information. To address these limitations, we introduce aKnowledge-Guided Graph (KGG) framework employingself-supervised learning to pretrain models using orbital-level featuresin order to mitigate reliance on extensive labeled data sets. In addition,we propose novel representations for atomic hybridization and bondtypes that explicitly consider orbital engagement. Our pretrainingstrategy is cost efficient, utilizing approximately 250,000 moleculesfrom the ZINC15 data set, in contrast to contemporary approaches thattypically require between two and ten million molecules, consequentlyreducing the risk of potential data contamination. Extensive evaluationson diverse downstream molecular property data sets demonstrate thatour method significantly outperforms state-of-the-art baselines. Complementaryanalyses, including t-SNE visualizations andcomparisons with traditional molecular fingerprints, further validatethe effectiveness and robustness of our proposed KGG approach. The key advantages of KGG are itsdata efficiency and architectural versatility, driven by orbital-informedrepresentations. By distilling essential chemical knowledge from modestcorpora, it avoids extensive pretraining and excels in low-data fine-tuning,providing a robust and chemically meaningful foundation for diverse GNN architectures.

Synthetic methods for unsymmetrical disulfides are greatly needed owing to their applications in drug discovery, linker chemistry, and materials sciences. In this study, a new shelf-stable and easy-to-prepare bilateral disulfurating platform molecule, N-(morpholine-4-dithio)phthalimide, has been developed for the divergent synthesis of unsymmetrical disulfides. The amino and imide leaving groups of this reagent can be orthogonally transformed. Under acidic conditions, the amino moiety undergoes selective protonation and thus can be displaced by various carbon nucleophiles, such as allyl trimethylsilanes, alkynyl silanes, and electron-rich arenes, leaving the phthalimide moiety untouched. Meanwhile, the phthalimide leaving group is amenable to substitution under basic or neutral conditions. The combination of these transformations provides rapid access to diverse unsymmetrical disulfides through two C-S bond-forming reactions.

This review explores recent advancements in synthesizing quinoid heteroaryls, namely quinazoline and quinoline, vital in chemistry due to their prevalence in natural products and pharmaceuticals. It emphasizes the rapid, highly efficient, and economically viable synthesis achieved through gold-catalyzed cascade protocols. By investigating methodologies and reaction pathways, the review underscores exceptional yields attainable in the synthesis of quinoid heteroaryls. It offers valuable insights into accessing these complex structures through efficient synthetic routes. Various strategies, including cyclization, heteroarylation, cycloisomerization, cyclo-condensation, intermolecular and intramolecular cascade reactions, are covered, highlighting the versatility of gold-catalyzed approaches. The comprehensive compilation of different synthetic approaches and elucidation of reaction mechanisms contribute to a deeper understanding of the field. This review paves the way for future advancements in synthesizing quinoid heteroaryls and their applications in drug discovery and materials science.

Multicomponent reactions, which involve the reaction of three or more components in a single step to efficiently produce a new molecular structure, have received growing attention in a range of research areas. In this study, four- and five-component reactions incorporating alkenes, carboxylic acids, alkyl radicals, and nitriles are developed for the imidoalkylation of alkenes with perfect regioselectivity. More specifically, this approach involves a photoredox reaction based on a catalytic indirect reductive quenching system. Various drug molecules and amino acids were demonstrated to be compatible substrates for this reaction, thereby enabling the synthesis of multifunctionalized derivatives. Notably, five individual components were assembled with perfect regioselectivity via the reaction of a radical species in the presence of a polarity-matched alkene. Mechanistic studies and density functional theory calculations suggested that this multicomponent reaction involves a cascade of Ritter-type substitutions and Mumm rearrangements. This method exhibits immense potential for use in the creation of structurally diverse and biologically relevant molecules with defined connectivities, thereby holding promise for applications in drug discovery and materials science.

Artificial intelligence (AI) has been transforming the practice of drug discovery in the past decade. Various AI techniques have been used in many drug discovery applications, such as virtual screening and drug design. In this survey, we first give an overview on drug discovery and discuss related applications, which can be reduced to two major tasks, i.e. molecular property prediction and molecule generation. We then present common data resources, molecule representations and benchmark platforms. As a major part of the survey, AI techniques are dissected into model architectures and learning paradigms. To reflect the technical development of AI in drug discovery over the years, the surveyed works are organized chronologically. We expect that this survey provides a comprehensive review on AI in drug discovery. We also provide a GitHub repository with a collection of papers (and codes, if applicable) as a learning resource, which is regularly updated.

This research introduces two novel topological indices, the entire Albertson index and the entire sigma index, as quantitative measures of molecular irregularity. The indices are defined by precise mathematical formulas and their behavior is analyzed across a diverse range of graph families. To evaluate the predictive capabilities of the proposed indices, we compare their performance with established irregularity indices in the modeling of molecular properties. Correlations with physicochemical properties, including the boiling point, melting point, and molecular volume, are investigated. Specific expressions for these indices are derived for various molecular structures, such as bridge molecules, polyomino chains of n-cycles, triangular benzenoid graphs, graphene, and dendrimer stars D3[n]. The findings of this study contribute significantly to the field of chemical graph theory by providing novel tools to understand and predict molecular behavior. The entire irregularity indices have potential applications in drug discovery, materials science, and other areas where molecular properties are crucial.

Triarylmethanes are important structural motifs with wide applications in natural products, drug discovery, and materials science. Chiral triarylmethanes are particularly noteworthy due to their unique three-dimensional architectures, which facilitate interactions with biological targets. Current enantioselective syntheses primarily rely on strategies such as enantioselective desymmetrization, asymmetric dehydroxylation, and stereocontrolled C-C bond formation. However, these methods often depend on preconstructed frameworks, posing challenges for synthesizing unfunctionalized triarylmethanes. Herein, we present a one-pot cascade reaction that enables the modular and enantioselective synthesis of triarylmethanes from aldehyde-derived hydrazones, aryl halides, and aryl nucleophiles. This method achieves excellent step and pot economies by simultaneously forming two distinct C-C bonds. Notably, the enantiodetermining step is strategically relocated to the transition metal-carbene migratory insertion event, overcoming limitations of conventional approaches. Our results demonstrate robust yields and excellent enantioselectivity across a diverse range of substrates. This modular strategy also allows for facile access to both enantiomers by simply switching the aromatic rings. DFT calculations reveal that the transmetalation step is the rate-determining step and the carbenation process is the enantioselectivity-determining step, also demonstrating the self-adaptive nature of the SadPhos ligand.

Providing explainable molecular property predictions is critical for many scientific domains, such as drug discovery and material science. Though transformer-based language models have shown great potential in accurate molecular property prediction, they neither provide chemically meaningful explanations nor faithfully reveal the molecular structure-property relationships. In this work, we develop a framework for explainable molecular property prediction based on language models, dubbed as Lamole, which can provide chemical concepts-aligned explanations. We take a string-based molecular representation - Group SELFIES - as input tokens to pre-train and fine-tune our Lamole, as it provides chemically meaningful semantics. By disentangling the information flows of Lamole, we propose considering both self-attention weights and gradients for better quantification of each chemically meaningful substructure's impact on the model's output. To make the explanations more faithful to the structure-property relationship, we then carefully craft a marginal loss to explicitly optimize the explanations to align with the chemists' annotations. We bridge the manifold hypothesis with the elaborated marginal loss to prove that the loss can align the explanations with the tangent space of the data manifold, leading to concept-aligned explanations. Experimental results over eight datasets demonstrate Lamole can achieve comparable prediction accuracy and boost the explanation accuracy by up to 14.3%, being the state-of-the-art in explainable molecular property prediction. To further illustrate the actionable utility of the explanations derived from Lamole, we integrated the framework with an evolutionary algorithm. This integration established an interpretable optimization pipeline for molecular editing, demonstrating that Lamole functions beyond simple post-hoc analysis but serves as a practical guide for molecule discovery.

Efficient molecular property prediction is crucial in bioinformatics and cheminformatics, with applications in drug discovery, materials science, and chemical engineering. This paper introduces MolHFCNet, a graph neural network designed to enhance molecular representation learning. At its core, the n-Hierarchical Features Combining (n-HFC) module aggregates information across multiple hierarchical feature spaces, effectively capturing both local and global graph structures. Unlike conventional models, n-HFC maintains computational complexity comparable to a single full-dimensional graph layer while supporting either 2D or 3D molecular graphs, ensuring flexibility across tasks. Furthermore, we propose a novel graph pretraining strategy that integrates predictive and contrastive learning, enabling the model to capture local chemical interactions and global molecular contexts for robust embeddings. Experimental results on benchmark datasets demonstrate MolHFCNet’s superior accuracy and efficiency compared to state-of-the-art methods, highlighting the potential of high-order hierarchical feature learning for advancing molecular graph analysis. Our code is available at https://github.com/ndlongvn/MolHFCNet.

Molecular docking is a key tool in structural biology and computer-assisted drug design. Molecular docking is a method which predicts the preferred orientation of a ligand when bound in an active site to form a stable complex. It is the most common method used as a structure-based drug design. Here, the authors intend to discuss the various types of docking methods and their development and applications in modern drug discovery. The important basic theories such as sampling algorithm and scoring functions have been discussed briefly. The performances of the different available docking software have also been discussed. This chapter also includes some application examples of docking studies in modern drug discovery such as targeted drug delivery using carbon nanotubes, docking of nucleic acids to find the binding modes and a comparative study between high-throughput screening and structure-based virtual screening.

The acid-base dissociation constant (pKa) characterizes a molecule's tendency to donate or accept protons, thereby fundamentally influencing its physicochemical properties and behavior. Consequently, pKa is pivotal to diverse applications in drug discovery, including virtual screening and drug design. However, accurate pKa prediction remains challenging, due to intricate atomic interactions among ionizable groups and the scarcity of experimental data. To address these, we propose TripKa, a quantum-informed aqueous pKa predictor that employs a triplet interaction network to model complex thermodynamic coupling in multisite protonation equilibria. TripKa is fine-tuned using quantum-level descriptors as training targets, enabling the extraction of physical insights into molecular acid-base dissociation properties from limited high-quality pKa data. Our proposed TripKa outperforms state-of-the-art methods on multiple pKa benchmarks, including the SAMPL6-8 and Novartis data sets, achieving reductions of over 10% in both mean absolute error and root-mean-square error. Beyond pKa prediction, TripKa further serves as a foundational model for diverse molecular properties prediction and interaction modeling tasks, benefiting from its chemically informed pKa-pretraining.

The application of multimodal image fusion has become increasingly widespread across various fields in the era of deep learning. Existing fusion methods integrate infrared and visible images to provide complementary content and enhance the robustness of complex real-world scenes for high-level visual tasks, such as semantic segmentation and object detection. In return, high-level visual tasks facilitate the fusion of infrared and visible by providing mid-level semantic information. However, such frameworks rely heavily on multimodal data and require strict registration of images from different modalities before fusion, seriously limiting their practical applications due to the common realistic situations of missing modalities or misregistration. To move beyond this limitation, we propose a novel hierarchical knowledge distillation (HKD) framework tailored for unimodal image segmentation with the guidance of multi-modality. This framework aims to retain as much diverse information from multimodal image fusion as possible, thereby enhancing downstream high-level visual tasks when only the unimodal images are available during the inference phase. Our proposed method is two-stage, and we construct a robust multimodal fusion and segmentation interaction network in the first stage as a powerful teacher model. In the second stage, we design a hierarchical distillation method to transfer the fused and segmented multi-layer knowledge from the multimodal teacher model to the unimodal student model. Extensive experimental results on two public datasets, i.e., MFNet and FMB, demonstrate that the proposed hierarchical knowledge distillation framework can effectively transfuse multimodal knowledge into the unimodal student model for image enhancement and segmentation under incomplete multimodal conditions, and achieves considerably competitive results compared to multimodal image fusion and segmentation models.

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.