Cheminformatics

As efforts to computationally describe and simulate the biochemical world become more commonplace, computer programs that are capable of in silico chemistry play an increasingly important role in biochemical research. While such programs exist, they are often dependency-heavy, difficult to navigate, or not written in Python, the programming language of choice for bioinformaticians. Here, we introduce PIKAChU (Python-based Informatics Kit for Analysis CHemical Units): a light-weight cheminformatics toolbox implemented in Python. PIKAChU builds comprehensive molecular graphs from SMILES strings, which allow for easy downstream analysis and visualisation of molecules. While the molecular graphs PIKAChU generates are extensive, storing and inferring information on aromaticity, chirality, charge, hybridisation and electron orbitals, PIKAChU limits itself to applications that will be sufficient for most casual users and downstream Python-based tools and databases, such as Morgan fingerpri...

The semiconductor manufacturing sector produces enormous amounts of textual data that is highly
imbalanced, non-stationary, and operationally critical. Although transformer-based language
models achieve strong classification accuracy, their robustness and probability calibration under
industrial constraints remain insufficiently addressed, particularly in resource-limited deployments.
This paper proposes LiteFormer, a lightweight and calibrated transformer framework for
imbalanced industrial text classification. The technique combines a geometry-informed minority
over-sampling technique with D-SMOTE, imbalance-informed optimization with Focal Loss, and a
post-hoc temperature scaling method. The technique outperforms standard transformer models on
a large-scale industrial Root Cause Analysis data set, obtaining higher macro-F1 and significantly
better Expected Calibration Error, while remaining computationally efficient. The technique
performs robustly even when faced with temporal and domain shifts.

Molecular analytics encompasses methods for property prediction, generation, and related tasks applicable across chemistry and biology. Deep learning (DL) methods based on neural architectures aim to leverage large datasets to produce high-quality representations and models. The scope is broad, including progressions through various types of molecular representations, datasets and evaluation, principal neural architectures, training and evaluation regimes, established applications, and key research challenges. The objective is to delineate core developments and to illuminate straightforward yet compelling research directions in molecular analytics. Diverse property prediction benchmarks have been proposed, notably including the Chemistry, Manufacturing, and Controls dataset for prediction of chemical manufacturing. Additional classical quantitative structure-activity relationship tasks propose endurance as an indicator of prediction reliability for poisoning or mutagenesis. Molecular generation and virtual screening have been closely coupled since the inception of these methods, and community efforts continue to refine the quality and capability of such pipelines. The intuitive goal of mapping molecular representations to chemical properties remains a central pursuit of molecular analytics; the best-performing methods draw inspiration from established chemistry and physics. Recent work has illustrated the potential of applying generative models of molecular surfaces in conjunction with virtual screens for next-generation screening of combinatorial libraries. 6.2. Foundations of Molecular Analytics The importance of molecular analytics is driven by fundamental aspirations in science and technology. Insights into biological systems and their interactions, from local to global scales, underpin advances in health and medicine; new materials tailored for

• Model developed using "public data" -May have low prediction accuracy due to limited chemical diversity of both sets (problem of accuracy) • Model developed using "public" + "in house" data -Difficult to receive "in house" data -"Public data" may not be really public (problem of IP) -May require significant computing time/expertise of the end user -Public/in-house data can be incompatible • Model developed using "public" data and then "adjusted" for molecules from "in house" data -Associative Neural Network  Some software tools rely just on one "best" model.  Other software tools rely on the ensemble average ("panel of experts").  ASNN explores disagreement of individual models in the ensemble to improve its accuracy and to derive a confidence score.

Essential oils (EOs) are popular in aromatherapy, a branch of alternative medicine that claims their curative effects. Moreover, several studies reported EOs as potential anti-cancer agents by inducing apoptosis in different cancer cell models. In this study, we have considered EOs as a potential resource of new kinase inhibitors with a polypharmacological profile. On the other hand, computational methods offer the possibility to predict the theoretical activity profile of ligands, discovering dangerous off-targets and/or synergistic effects due to the potential multi-target action. With this aim, we performed a Structure-Based Virtual Screening (SBVS) against X-ray models of several protein kinases selected from the Protein Data Bank (PDB) by using a chemoinformatics database of EOs. By evaluating theoretical binding affinity, 13 molecules were detected among EOs as new potential kinase inhibitors with a multi-target profile. The two compounds with higher percentages in the EOs wer...

Background: Three-dimensional (3D) imaging mass spectrometry (MS) is an analytical chemistry technique for the 3D molecular analysis of a tissue specimen, entire organ, or microbial colonies on an agar plate. 3D-imaging MS has unique advantages over existing 3D imaging techniques, offers novel perspectives for understanding the spatial organization of biological processes, and has growing potential to be introduced into routine use in both biology and medicine. Owing to the sheer quantity of data generated, the visualization, analysis, and interpretation of 3D imaging MS data remain a significant challenge. Bioinformatics research in this field is hampered by the lack of publicly available benchmark datasets needed to evaluate and compare algorithms. Findings: High-quality 3D imaging MS datasets from different biological systems at several labs were acquired, supplied with overview images and scripts demonstrating how to read them, and deposited into MetaboLights, an open repository for metabolomics data. 3D imaging MS data were collected from five samples using two types of 3D imaging MS. 3D matrix-assisted laser desorption/ionization imaging (MALDI) MS data were collected from murine pancreas, murine kidney, human oral squamous cell carcinoma, and interacting microbial colonies cultured in Petri dishes. 3D desorption electrospray ionization (DESI) imaging MS data were collected from a human colorectal adenocarcinoma. Conclusions: With the aim to stimulate computational research in the field of computational 3D imaging MS, selected high-quality 3D imaging MS datasets are provided that could be used by algorithm developers as benchmark datasets.

The therapeutic potential of phytochemicals lies in their ability to modulate multiple diseaserelevant targets simultaneously, yet conventional reductionist assays often fail to capture this polypharmacology. Network pharmacology, when integrated with in silico methods such as molecular docking, molecular dynamics simulations, and cheminformatics, offers a powerful systems-level framework to decode the multi-target mechanisms of plant-derived compounds. This review critically evaluates the state of the art, comparing computational tools, databases, and validation strategies. I synthesize case studies across inflammation, cancer, and neurodegeneration to identify best practices and persistent pitfalls, including overreliance on binding affinity cutoffs, neglect of pharmacokinetic constraints, and insufficient experimental validation. Emerging solutions, such as machine learning-based target prediction, pharmacophore-constrained network analysis, and integrated ADMET filtering, are discussed. I conclude that the future of phytochemical network pharmacology lies in quantitative, predictive models that prioritize functional network perturbation over simple topological metrics, coupled with standardized experimental validation pipelines.

Chitosan (CS) and its derivatives are important particles for the administration of nanomedicine and drugs. Using bioinformatics and cheminformatics tools, the molecular descriptors of chitosan and other new derivatives were quantified and compared using the nano-SAR method.

Bartonella bacilliformis is the causative agent of a neglected tropical disease, the Oroya fever or Carrion's disease. Current treatment options for B. bacilliformis infections are limited, and the emergence of antibiotic-resistant strains underscores the urgent need for novel therapeutic interventions. This study aimed to identify novel inhibitors from Tibetan medicinal plant metabolites against B. bacilliformis, a causative agent of Oroya fever, utilizing a multi-pronged computational approach. We employed pan-genomics and hierarchal subtractive proteomics approach to filter drug targets and then predicted binding affinities of compounds with riboflavin synthase target using AutoDock Vina, DiffDock, and GNINA. The stability of the top inhibitors was confirmed through molecular dynamics (MD) simulations in GROMACS, while MM/GBSA and ADMET profiling assessed pharmacokinetic and toxicity properties. The analysis identified Kaempferol 3-O-gentiobioside, Kaempferol 3-(6''-p-coumarylglucoside)-7-glucoside, and Hirtusneanoside as promising inhibitors. MD simulations demonstrated stable binding with RMSD within 3 Å, and the candidates showed no skin sensitization, AMES mutagenicity, or hepatotoxicity. Physiologically-based pharmacokinetic (PBPK) modeling estimated compound distribution within the human body and revealed slightly higher absorption in pregnant and steatosis conditions, but overall absorption rates were similar across scenarios. Cyclodextrin complexation, particularly with SBE-β-CD indicates potential for enhanced solubility of these ligands. Since Kaempferol 3-O-gentiobioside exhibited off-target effects, the Kaempferol 3-(6''-p-coumarylglucoside)-7-glucoside and Hirtusneanoside are recommended for further investigation. Further testing in laboratory is recommended to explore these leads options for treating Oroya fever.

A variety of software packages are available for the combinatorial enumeration of virtual libraries for small molecules, starting from specifications of core scaffolds with attachments points and lists of R-groups as SMILES or SD files. Although SD files include atomic coordinates for core scaffolds and R-groups, it is not possible to control 2-dimensional (2D) layout of the enumerated structures generated for virtual compound libraries because different packages generate different 2D representations for the same structure. We have developed a software package called LipidMapsTools for the template-based combinatorial enumeration of virtual compound libraries for lipids. Virtual libraries are enumerated for the specified lipid abbreviations using matching lists of pre-defined templates and chain abbreviations, instead of core scaffolds and lists of R-groups provided by the user. 2D structures of the enumerated lipids are drawn in a specific and consistent fashion adhering to the framework for representing lipid structures proposed by the LIPID MAPS consortium. LipidMapsTools is lightweight, relatively fast and contains no external dependencies. It is an open source package and freely available under the terms of the modified BSD license.

A variety of software packages are available for the combinatorial enumeration of virtual libraries for small molecules, starting from specifications of core scaffolds with attachments points and lists of R-groups as SMILES or SD files. Although SD files include atomic coordinates for core scaffolds and R-groups, it is not possible to control 2-dimensional (2D) layout of the enumerated structures generated for virtual compound libraries because different packages generate different 2D representations for the same structure. We have developed a software package called LipidMapsTools for the template-based combinatorial enumeration of virtual compound libraries for lipids. Virtual libraries are enumerated for the specified lipid abbreviations using matching lists of pre-defined templates and chain abbreviations, instead of core scaffolds and lists of R-groups provided by the user. 2D structures of the enumerated lipids are drawn in a specific and consistent fashion adhering to the fra...

The integration of artificial intelligence (AI) in pharmaceutical research has accelerated drug discovery processes, particularly through predictive modeling of drug mechanisms. Traditional supervised learning approaches rely heavily on large, labeled datasets, which are often expensive and time-consuming to curate. Self-supervised learning (SSL) offers a promising alternative by enabling models to learn meaningful representations from unlabeled molecular and biochemical data, thereby reducing dependency on annotated datasets. This study explores the application of self-supervised AI models for predicting drug mechanisms, including molecular interactions, target engagement, and pharmacodynamic effects. By leveraging graph-based representations of molecules, protein interaction networks, and large-scale omics datasets, SSL models can uncover latent biological patterns that inform mechanistic predictions. Comparative analyses with conventional supervised models demonstrate that self-supervised models achieve competitive or superior predictive performance while requiring fewer labeled examples. Additionally, SSL approaches enhance model generalizability across diverse chemical scaffolds and therapeutic classes, facilitating rapid screening and repurposing of drug candidates. The study also addresses challenges such as data heterogeneity, interpretability of learned representations, and integration with downstream experimental validation pipelines. Overall, this work underscores the transformative potential of self-supervised AI in pharmaceutical research, highlighting its role in accelerating drug discovery, optimizing preclinical workflows, and supporting mechanistic understanding of drug actions.

Tumor necrosis factor-α (TNF-α) is a drug target in rheumatoid arthritis and several other auto-immune disorders. TNF-α binds with TNF receptors (TNFR), located on the surface of several immunological cells to exert its effect. Hence, the use of inhibitors that can hinder the complex formation of TNF-α/TNFR can be of medicinal significance. In this study, multiple chem-informatics approaches, including descriptor-based screening, 2D-similarity searching, and pharmacophore modelling were applied to screen new TNF-α inhibitors. Subsequently, multiple-docking protocols were used, and four-fold post-docking results were analyzed by consensus approach. After structurebased virtual screening, seventeen compounds were mutually ranked in top-ranked position by all the docking programs. Those identified hits target TNF-α dimer and effectively block TNF-α/TNFR interface. The predicted pharmacokinetics and physiological properties of the selected hits revealed that, out of seventeen, seven compounds (4, 5, 10, 11, 13-15) possessed excellent ADMET profile. These seven compounds plus three more molecules (7, 8 and 9) were chosen for molecular dynamics simulation studies to probe into ligand-induced structural and dynamic behavior of TNF-α, followed by ligand-TNF-α binding free energy calculation using MM-PBSA. The MM-PBSA calculations revealed that compounds 4, 5, 7 and 9 possess highest affinity for TNF-α; 8, 11, 13-15 exhibited moderate affinities, while compound 10 showed weaker binding affinity with TNF-α. This study provides valuable insights to design more potent and selective inhibitors of TNF-α, that will help to treat inflammatory disorders.

Tumor necrosis factor-α (TNF-α) is a drug target in rheumatoid arthritis and several other auto-immune disorders. TNF-α binds with TNF receptors (TNFR), located on the surface of several immunological cells to exert its effect. Hence, the use of inhibitors that can hinder the complex formation of TNF-α/TNFR can be of medicinal significance. In this study, multiple chem-informatics approaches, including descriptor-based screening, 2D-similarity searching, and pharmacophore modelling were applied to screen new TNF-α inhibitors. Subsequently, multiple-docking protocols were used, and four-fold post-docking results were analyzed by consensus approach. After structurebased virtual screening, seventeen compounds were mutually ranked in top-ranked position by all the docking programs. Those identified hits target TNF-α dimer and effectively block TNF-α/TNFR interface. The predicted pharmacokinetics and physiological properties of the selected hits revealed that, out of seventeen, seven compounds (4, 5, 10, 11, 13-15) possessed excellent ADMET profile. These seven compounds plus three more molecules (7, 8 and 9) were chosen for molecular dynamics simulation studies to probe into ligand-induced structural and dynamic behavior of TNF-α, followed by ligand-TNF-α binding free energy calculation using MM-PBSA. The MM-PBSA calculations revealed that compounds 4, 5, 7 and 9 possess highest affinity for TNF-α; 8, 11, 13-15 exhibited moderate affinities, while compound 10 showed weaker binding affinity with TNF-α. This study provides valuable insights to design more potent and selective inhibitors of TNF-α, that will help to treat inflammatory disorders.

Tumor necrosis factor-α (TNF-α) is a drug target in rheumatoid arthritis and several other auto-immune disorders. TNF-α binds with TNF receptors (TNFR), located on the surface of several immunological cells to exert its effect. Hence, the use of inhibitors that can hinder the complex formation of TNF-α/TNFR can be of medicinal significance. In this study, multiple chem-informatics approaches, including descriptor-based screening, 2D-similarity searching, and pharmacophore modelling were applied to screen new TNF-α inhibitors. Subsequently, multiple-docking protocols were used, and four-fold post-docking results were analyzed by consensus approach. After structurebased virtual screening, seventeen compounds were mutually ranked in top-ranked position by all the docking programs. Those identified hits target TNF-α dimer and effectively block TNF-α/TNFR interface. The predicted pharmacokinetics and physiological properties of the selected hits revealed that, out of seventeen, seven compounds (4, 5, 10, 11, 13-15) possessed excellent ADMET profile. These seven compounds plus three more molecules (7, 8 and 9) were chosen for molecular dynamics simulation studies to probe into ligand-induced structural and dynamic behavior of TNF-α, followed by ligand-TNF-α binding free energy calculation using MM-PBSA. The MM-PBSA calculations revealed that compounds 4, 5, 7 and 9 possess highest affinity for TNF-α; 8, 11, 13-15 exhibited moderate affinities, while compound 10 showed weaker binding affinity with TNF-α. This study provides valuable insights to design more potent and selective inhibitors of TNF-α, that will help to treat inflammatory disorders.

Cardiobacterium valvarum is a known cause of infective endocarditis (IE). Autoimmune responses have been implicated in contributing to the inflammatory processes observed in IE. Molecular mimicry significantly influences the interaction between pathogens and host immune systems, frequently resulting in autoimmune responses. This study employs a bioinformatics approach to identify candidate proteins exhibiting molecular mimicry in C. valvarum that might elicit immune reactions and contribute to inflammation. A total of 18 proteins depicted potential for autoimmunity based on homology and antigenicity. Further, subtractive proteomics was then applied to identify potential drug targets in the pathogen, and purK gene product (5-(carboxyamino)imidazole ribonucleotide synthase) was utilized for downstream analysis. Molecular docking was used to assess the inhibitory potential of natural products derived from Ocimum tenuiflorum, Leonurus cardiaca, and Panax ginseng, which have previously been associated with protective effects against endocarditis. Five compounds were prioritized based on their AutoDock Vina binding scores: (1) a complex triterpenoid from P. ginseng, (2) a flavonoid derivative from L. cardiaca, and (3) a steroid and flavonoid from O. tenuiflorum. Among these, LTS0201798 and LTS0158828 appear to be more suitable for oral administration, while using a nanocarrier or complexing with cyclodextrin could enhance the suitability of the other compounds as well. LTS0201798 and LTS0158828 also showed good binding affinity and stable complexation over 100 ns of molecular dynamics simulation. Hence, we propose that future research should focus on validating our findings through experimental assays to confirm the efficacy of these compounds in clinical settings.

The purpose of High Throughput Screening (HTS) in pharmaceutical industry is to identify, as soon as possible, compounds that are good starting points for successful new drug development process. Experts from this area study the chemical structures of so called »hit« compounds that have been found to interact with the target protein, interfere with proliferation of different types of cells or stop bacterial or fungal growth. Hypotheses to design related structures with improved biological properties are than builded. Each idea is then tested by the iterative synthesis and testing of novel compounds in various biological assays, searching for hits with better properties and defining useful and promising »lead« molecules. In parallel, molecular modeling and chemoinformatics experts can increase efficiency and decrease experimental costs by using different database filtering methods. In such a way, hits from HTS may be assessed before committing significant resource for chemical optimi...

Molecular Property Diagnostic Suite (MPDS) was conceived and developed as an open-source disease-specific web portal based on Galaxy. MPDS COVID-19 was developed for COVID-19 as a onestop solution for drug discovery research. Galaxy platforms enable the creation of customized workflows connecting various modules in the web server. The architecture of MPDS COVID-19 effectively employs Galaxy v22.04 features, which are ported on CentOS 7.8 and Python 3.7. MPDS COVID-19 provides significant updates and the addition of several new tools updated after six years. Tools developed by our group in Perl/Python and open-source tools are collated and integrated into MPDS COVID-19 using XML scripts. Our MPDS suite aims to facilitate transparent and open innovation. This approach significantly helps bring inclusiveness in the community while promoting free access and participation in software development.

This study investigates the implementation of Blender 3D software as an educational tool for creating visual scientific content by chemistry students. We evaluated undergraduate and graduate students (n = 15) through precourse questionnaires for constructing technological acceptance profiles and postcourse questionnaires based on the Technology Acceptance Model (TAM), in addition to analyzing their creation portfolios. The results reveal that although most participants (11/15) had no prior experience and perceived the software as complex, the high perceived usefulness of the tool for their academic objectives superseded the low perceived ease of use. Guided by the intervention, students progressed from modeling laboratory equipment (macroscopic) to creating complex molecular visualizations (submicroscopic), such as electrochemical cells and lipid micelles. We conclude that success in tool adoption was dependent on the pedagogical support structure, including a technical manual and a progression of scaffolded tasks.

MAOs are flavoenzymes that aid in the oxidative deamination of neurotransmitters such as dopamine, serotonin, and epinephrine. MAO inhibitors are antidepressants that act by inhibiting neurotransmitter breakdown in the brain and controlling mood. MAO inhibitors with the chlorophenyl-chromone-carboxamide structure have been shown in investigations to be extremely effective. The current study employs in-silico screening, MD simulation, and drug kinetics evaluation, all of which are evaluated using different criteria. The study comprised 37 ligands, and three stood out as the best, with greater binding scores above the threshold value. Docking analysis found that compound 34 had the highest docking score in the series (À13.60 kcal/mol) and interacted with the important amino acids TYR 435, CYS 397, CYS 172, PHE 343, TYR 398, and LYS 296 required for MAO inhibitory activity. The ADMET study revealed that the compounds had drug-like properties. The results of this study could be used to develop chromone drugs that target the MAO inhibitor. The top three ligands with the highest force and work were then simulated using molecular dynamics. The protein-ligand complexes had steady trajectories throughout the 100 ns simulation, according to the data. Furthermore, the drug likeliness predicted by ADMET analysis findings indicated that the top three lead compounds had strong inhibitory efficiency, superior pharmacokinetics, and were non-toxic under physiological settings. As a result, these compounds have the potential to be exploited as possible treatment medications for PD.

Comparative protein modeling of a target protein based on sequence similarity to a protein with known structure is widely used to provide structural models of proteins. Frequently, the quality of the target- template sequence alignment is non-uniform along the sequence: parts can be modeled with a high confidence, whereas other parts differ strongly from the template. In principle, molecular dynamics (MD) simulations can be used to refine protein model structures and also to model loops in homology modeled protein structures, but it is limited by the currently accessible simulation time scales. In the current work we have used a recently developed biasing potential replica exchange (BP-Rex) MD [1] method to refine and to model loops in homology modeled protein structure at atomic resolution including explicit solvent. In standard Rex MD [2] simulations several replicas of a system are run in parallel at different temperatures allowing exchanges at preset time intervals. In a BP-Rex ...

The maximum common property similarity (MCPhd) method is presented using descriptors as a new approach to determine the similarity between two chemical compounds or molecular graphs. This method uses the concept of maximum common property arising from the concept of maximum common substructure and is based on the electrotopographic state index for atoms. A new algorithm to quantify the similarity values of chemical structures based on the presented maximum common property concept is also developed in this paper. To verify the validity of this approach, the similarity of a sample of compounds with antimalarial activity is calculated and compared with the results obtained by four different similarity methods: the small molecule subgraph detector (SMSD), molecular fingerprint based (OBabel_FP2), ISIDA descriptors and shape-feature similarity (SHAFTS). The results obtained by the MCPhd method differ significantly from those obtained by the compared methods, improving the quantification ...

Natural Products (NP) are essential for the discovery of novel drugs and products for numerous biotechnological applications. The NP discovery process is expensive and time-consuming, having as major hurdles dereplication (early identification of known compounds) and structure elucidation, particularly the determination of the absolute configuration of metabolites with stereogenic centers. This review comprehensively focuses on recent technological and instrumental advances, highlighting the development of methods that alleviate these obstacles, paving the way for accelerating NP discovery towards biotechnological applications. Herein, we emphasize the most innovative high-throughput tools and methods for advancing bioactivity screening, NP chemical analysis, dereplication, metabolite profiling, metabolomics, genome sequencing and/or genomics approaches, databases, bioinformatics, chemoinformatics, and three-dimensional NP structure elucidation.

Computer-based de-novo design of functional molecules is one of the most prominent challenges in cheminformatics today. As a result, generative and evolutionary inverse designs from the field of artificial intelligence have emerged at a rapid pace, with aims to optimize molecules for a particular chemical property. These models ‘indirectly’ explore the chemical space; by learning latent spaces, policies, and distributions, or by applying mutations on populations of molecules. However, the recent development of the SELFIES (Krenn 2020 Mach. Learn.: Sci. Technol. 1 045024) string representation of molecules, a surjective alternative to SMILES, have made possible other potential techniques. Based on SELFIES, we therefore propose PASITHEA, a direct gradient-based molecule optimization that applies inceptionism (Mordvintsev 2015) techniques from computer vision. PASITHEA exploits the use of gradients by directly reversing the learning process of a neural network, which is trained to pred...

Computer-based de-novo design of functional molecules is one of the most prominent challenges in cheminformatics today. As a result, generative and evolutionary inverse designs from the field of artificial intelligence have emerged at a rapid pace, with aims to optimize molecules for a particular chemical property. These models 'indirectly' explore the chemical space; by learning latent spaces, policies, distributions or by applying mutations on populations of molecules. However, the recent development of the SELFIES [1] string representation of molecules, a surjective alternative to SMILES, have made possible other potential techniques. Based on SELFIES, we therefore propose PASITHEA, a direct gradient-based molecule optimization that applies inceptionism [2] techniques from computer vision. PASITHEA exploits the use of gradients by directly reversing the learning process of a neural network, which is trained to predict real-valued chemical properties. Effectively, this forms an inverse regression model, which is capable of generating molecular variants optimized for a certain property. Although our results are preliminary, we observe a shift in distribution of a chosen property during inverse-training, a clear indication of PASITHEA's viability. A striking property of inceptionism is that we can directly probe the model's understanding of the chemical space it was trained on. We expect that extending PASITHEA to larger datasets, molecules and more complex properties will lead to advances in the design of new functional molecules as well as the interpretation and explanation of machine learning models.

Methods of computational quantum chemistry provide accurate approximations of molecular properties crucial for computer-aided drug discovery and other areas of chemical science. However, high computational complexity limits the scalability of their applications. Neural network potentials (NNPs) are a promising alternative to quantum chemistry methods, but they require large and diverse datasets for training. This work presents a new dataset and benchmark called ∇ 2 DFT that is based on the nablaDFT. It contains twice as much molecular structures, three times more conformations, new data types and tasks, and state-of-the-art models. The dataset includes energies, forces, 17 molecular properties, Hamiltonian and overlap matrices, and a wavefunction object. All calculations were performed at the DFT level (ωB97X-D/def2-SVP) for each conformation. Moreover, ∇ 2 DFT is the first dataset that contains relaxation trajectories for a substantial number of drug-like molecules. We also introduce a novel benchmark for evaluating NNPs in molecular property prediction, Hamiltonian prediction, and conformational optimization tasks. Finally, we propose an extendable framework for training NNPs and implement 10 models within it.

The electron density is an important object in quantum chemistry that is crucial for many downstream tasks in drug design. Recent deep learning approaches predict the electron density around a molecule from atom types and atom positions. Most of these methods use the plane wave (PW) numerical method as a source of ground-truth training data. However, the drug design field mostly uses the Linear Combination of Atomic Orbitals (LCAO) for computation of quantum properties. In this study, we focus on prediction of the electron density for drug-like substances and training neural networks with LCAO-based datasets. Our experiments show that proper handling of large amplitudes of core orbitals is crucial for training on LCAO-based data. We propose to store the electron density with the standard grids instead of the uniform grid. This allowed us to reduce the number of probing points per molecule by 43 times and reduce storage space requirements by 8 times. Finally, we propose a novel architecture based on the Deep-DFT model that we name LAGNet. It is specifically designed and tuned for drug-like substances and ∇ 2 DFT dataset. Scientific contribution We propose a core suppression model to correctly handle core orbitals and train neural network on LCAO-based data with atoms of the 3rd and 4th periods. We show that using the standard grid instead of the uniform grid drastically reduces the number of electron density probing points and data storage requirements. Finally, we propose the LAGNet model that allows to get better results on drug-like substances than the equivariant DeepDFT model.

Natural products with polypharmacological profiles have demonstrated promise as novel therapeutics for various complex diseases, including cancer. Currently, many gaps exist in our knowledge of which compounds interact with which targets, and experimentally testing all possible interactions is infeasible. Recent advances and developments of systems pharmacology and computational (in silico) approaches provide powerful tools for exploring the polypharmacological profiles of natural products. In this review, we introduce recent progresses and advances of computational tools and systems pharmacology approaches for identifying drug targets of natural products by focusing on the development of targeted cancer therapy. We survey the polypharmacological and systems immunology profiles of five representative natural products that are being considered as cancer therapies. We summarize various chemoinformatics, bioinformatics and systems biology resources for reconstructing drug-target networks of natural products. We then review currently available computational approaches and tools for prediction of drug-target interactions by focusing on five domains: target-based, ligand-based, chemogenomicsbased, network-based and omics-based systems biology approaches. In addition, we describe a practical example of the application of systems pharmacology approaches by integrating the polypharmacology of natural products and large-scale cancer genomics data for the development of precision oncology under the systems biology framework. Finally, we highlight the promise of cancer immunotherapies and combination therapies that target tumor ecosystems (e.g. clones or 'selfish' subclones) via exploiting the immunological and inflammatory 'side' effects of natural products in the cancer post-genomics era.

Identification of drug-target interaction (DTI) is an important challenge for research and development in the pharmaceutical industry. Biomedicine researchers have stepped from in vitro and in vivo experiments to in-silico methods for fast results. In the recent past, machine learning algorithms have become very popular for DTI predictions. This paper presents an ensemble approach- Random forest algorithm for DTI predictions. The performance of proposed approach is evaluated with respect to Matrix factorization, genetic algorithm, Support vector machines, K-nearest neighbor, Decision Trees and Logistic Regression over 4 benchmark datasets with diverse properties. The algorithm is evaluated over Accuracy and average ranking. Results establish that random forest algorithm is more suitable or DTI predictions as compared to other algorithms.

Sirtuin 1 (SIRT1) enzyme regulates major cell activities, and its activation offers lucrative therapeutic potentials for aging diseases including Alzheimer's disease (AD). Regarding the global aging society, continual attention has been given to various chemical scaffolds as a source for the discovery of novel SIRT1 activators since the discovery of the pioneer activator, resveratrol. Understanding structure-activity relationship (SAR) is essential for screening, designing as well as improving the properties of drugs. In this study, an in silico approach based on quantitative structure-activity relationship (QSAR) modeling, was employed for understanding the SAR of currently available SIRT1 fused-aromatic activators (i.e., imidazothiazole, oxazolopyridine, and azabenzimidazole analogs). Three QSAR models constructed using multiple linear regression (MLR) provided good predictive performance (R2LOOCV = 0.729 - 0.863 and RMSELOOCV = 0.165 - 0.325). An additional novel set of 181 s...

What makes a knowledge resource, like a domain model, thesaurus, or ontology, effective for term-level text annotation in the Biology domain? In this work we compare several approaches to ontology design with examples from well-known resources such as OBO, MeSH, and Genia. Based on these comparisons we establish goals for a knowledge resource that supports term-level text annotation and text mining: such a resource should represent terms and relations that are expressed in contiguous spans of text, and its terms should bear meaningful correspondences with other knowledge resources. Finally, we trace how these two goals have affected the re-design of the Genia Ontology over several iterations. The result is a new term hierarchy and a new design process, both specifically tailored to term-level text annotation. This research explores practical influences on the design of knowledge resources for Bio-NLP systems.

Quantitative structure activity relationships (QSAR) modelling is a well-known computational tool, often used in a wide variety of applications. Yet one of the major drawbacks of conventional QSAR modelling tools is that models are set up based on a limited number of experimental and/or theoretical conditions. To overcome this, the so-called multitasking or multi-target QSAR (mt-QSAR) approaches have emerged as new computational tools able to integrate diverse chemical and biological data into a single model equation, thus extending and improving the reliability of this type of modelling. We have developed QSAR-Co-X , an open source python−based toolkit (available to download at https://github.com/ncordeirfcup/QSAR-Co-X) for supporting mt-QSAR modelling following the Box-Jenkins moving average approach. The new toolkit embodies several functionalities for dataset selection and curation plus computation of descriptors, for setting up linear and non-linear models, as well as for a com...

Developing models able to predict interactions between drugs and enzymes is a primary goal in computational biology since these models may be used for predicting both new active drugs and the interactions between known drugs on untested targets. With the compilation of a large dataset of drug–enzyme pairs (62,524), we recognized a unique opportunity to attempt to build a novel multi-target machine learning (MTML) quantitative structure-activity relationship (QSAR) model for probing interactions among different drugs and enzyme targets. To this end, this paper presents an MTML-QSAR model based on using the features of topological drugs together with the artificial neural network (ANN) multi-layer perceptron (MLP). Validation of the final best model found was carried out by internal cross-validation statistics and other relevant diagnostic statistical parameters. The overall accuracy of the derived model was found to be higher than 96%. Finally, to maximize the diffusion of this model...

Deep eutectic solvents (DES) are often regarded as greener sustainable alternative solvents and are currently employed in many industrial applications on a large scale. Bearing in mind the industrial importance of DES—and because the vast majority of DES has yet to be synthesized—the development of cheminformatic models and tools efficiently profiling their density becomes essential. In this work, after rigorous validation, quantitative structure-property relationship (QSPR) models were proposed for use in estimating the density of a wide variety of DES. These models were based on a modelling dataset previously employed for constructing thermodynamic models for the same endpoint. The best QSPR models were robust and sound, performing well on an external validation set (set up with recently reported experimental density data of DES). Furthermore, the results revealed structural features that could play crucial roles in ruling DES density. Then, intelligent consensus prediction was em...

Brazil is one of the countries with the highest level of drug consumption in the world. By 2012 about 66% claimed to practice self-medication. Such activity can lead to a wide range of risks, including death from drug intoxication. Studies indicate that a lack of knowledge about drugs and their dangers is one of the main aggravating factors in this scenario. This work aims to universalize access to information about medications and their risks for different user profiles, especially Brazilian and lay users. In this paper, we presented the construction process of a Linked Data Mashup (LDM) integrating the datasets: consumer drug prices, government drug prices and drug’s risks in pregnant from ANVISA and SIDER from BIO2RDF. In addition, this work presents MediBot, an ontology-based chatbot capable of responding to requests in natural language in Portuguese through the instant messenger Telegram, smoothing the process to query the data. MediBot acts like a native language query interfa...

Cross‐linked block copolymers are structurally complex, and utilization of traditional methods of molecular representation in chemoinformatics is only of limited applicability. Therefore, we introduced new techniques of structural representation for block copolymers. We developed additive and combinatorial approaches that treat a copolymer as a mixture system. In this approach, DRAGON descriptors are concentration‐weighted for all chemicals in the reaction mixture. As a proof of concept, we have studied glass transition temperatures of block copolymers of hydroxyalkyl‐ and dihydroxyalkyl carbamate terminated poly(dimethylsiloxane) oligomers with poly(‐caprolactone) and developed four quantitative structure‐property relationships (QSPR) models. The correlation coefficient (R2) for mentioned QSPR models ranges from 0.851 to 0.911 for the training set. In addition to the newly introduced technique we found that the octanol−water partition coefficient and 3D‐MoRSE unweighted descriptors...

TidyGEO is a Web-based tool for downloading, tidying, and reformatting data series from Gene Expression Omnibus (GEO). As a freely accessible repository with data from over 4 million biological samples across more than 4,000 organisms, GEO provides diverse opportunities for secondary research. Transcriptomic data are most common in GEO, but other measurement types are also prevalent, including DNA methylation levels, genotypes, and chromatin-accessibility profiles. GEO's diversity and expansiveness present opportunities and challenges. Although scientists may find assay data relevant to a given research question, most analyses require sample annotations, such as a sample's treatment group, disease subtype, or age. In GEO, such annotations are stored alongside assay data in delimited, text-based files. However, the structure and semantics of the annotations vary widely from one series to another, and many annotations are not useful for analysis purposes. Thus, every GEO series must be tidied before it can be analyzed. Manual approaches may be used, but these are error prone and take time away from other research tasks. Custom computer scripts can be written, but many scientists lack the computational expertise to create such scripts. To address these challenges, we created TidyGEO, which supports essential data-cleaning tasks for sample-level annotations, such as selecting informative columns, .

Shallow machine learning methods have been applied to chemoinformatics problems with some success. As more data becomes available and more complex problems are tackled, deep machine learning methods may also become useful. Here we present a brief overview of deep learning methods and show in particular how recursive neural network approaches can be applied to the problem of predicting molecular properties. However molecules are typically described by undirected cyclic graphs, while recursive approaches typically use directed acyclic graphs. Thus we develop methods to address this discrepancy, essentially by considering an ensemble of recursive neural networks associated with all possible vertex-centered acyclic orientations of the molecular graph. One advantage of this approach is that it relies only minimally on the identification of suitable molecular descriptors, since suitable representations are learnt automatically from the data. Several variants of this approach are applied to the problem of predicting aqueous solubility and tested on four benchmark datasets. Experimental results show that the performance of the deep learning methods matches or exceeds the performance of other state-of-the-art methods according to several evaluation metrics and expose the fundamental limitations arising from training sets that are too small or too noisy. A web-based predictor AquaSol is available online through the ChemDB portal (cdb.ics.uci.edu) together with additional material.

Synthetic biology and metabolic engineering rely on computational search tools for predictions of novel biosynthetic pathways to industrially important compounds, many of which are derived from aromatic amino acids. Pathway search tools vary in their scope of covered reactions and compounds, as well as in metrics for ranking and evaluation. In this work, we present a new computational resource called ARBRE: Aromatic compounds RetroBiosynthesis Repository and Explorer. It consists of a comprehensive biochemical reaction network centered around aromatic amino acid biosynthesis and a computational toolbox for navigating this network. ARBRE encompasses over 28’000 known and 100’000 novel reactions predicted with generalized enzymatic reactions rules and over 70’000 compounds, of which 22’000 are known to biochemical databases and 48’000 only to PubChem. Over 1,000 molecules that were solely part of the PubChem database before and were previously impossible to integrate into a biochemica...

A classical virtual combinatorial chemistry approach (CombiChem) was applied for combinatorial generation of 5590 novel structurally-similar 6-fluoroquinolone analogs by using a virtual synthetic pathway with selected primary (43) and secondary amines (130). The obtained virtual combinatorial library was filtered using an in-house developed set of cheminformatics drug-likeness filters with pre-integrated Boolean options (TRUE/FALSE) for compounds reduction/selection. The retained number (304) of fluoroquinolone analogs (with TRUE outcome) defines the drug-like chemical space (CombiData). Quantitative structure-activity relationships (QSAR) study on these 304 virtually generated 6-fluoroquinolone analogs with unknown activity values was performed using a pre-built five-parameter multiple linear regression (MLR) model developed on a set of compounds with experimentally determined activity values (R tr = 0.8417, R tr-cv = 0.7884). The obtained activity values for the unknown compounds together with the model results were used to define the applicability domain (AD). The obtained AD offers a good graphical representation and establishment of structure-activity relationships (SAR) which could be used for design of new 6-fluoroquinolones with possible better activity.

Due to the lack of structural guidelines about G-quadruplex ligands, rational design cannot be the only approach to discover potent G4-ligands. As a complementary approach, screening of chemical library may provide interesting scaffolds known as hits provided that specific tools are available. In this work, the Institut Curie-CNRS chemical library was firstly screened by chemoinformatics methods. Similarity estimations by comparison with reference compounds (Phen-DC3, 360A, MMQ 12 ) provided a set of molecules, which were then evaluated by highthroughput G4-FID (HT-G4-FID) against various G-quadruplex DNA. A full investigation of the most interesting molecules, using the HT-G4-FID assay and molecular modeling, supplied an interesting structure-activity relationship confirming the efficiency of this general approach. Overall, we demonstrated that HT-G4-FID coupled with screening of chemical libraries is a powerful tool to identify new G4-DNA binding scaffolds.

When the above article was first published, a sentence on page 787 read as follows: "The AG Chemical Weapons Precursors list comprises a total of 87 dual-use chemicals that can be used as precursors for the synthesis of chemical weapons. Some of these precursors are also covered by the CWC Schedules, while other are not. All of the 87 entries in the AG list are explicitly listed as discrete chemicals [14, 23]." This sentence incorrectly characterized the Australia Group (AG) Chemical Weapons Precursors list as encompassing 87 chemicals. In fact, it encompasses 89 chemicals. The sentence has now been updated as follows in the online version of the article: "The AG Chemical Weapons Precursors list comprises a total of 89 dual-use chemicals that can be used as precursors for the synthesis of chemical weapons. Some of these precursors are also covered by the CWC Schedules, while other are not. All of the 89 entries in the AG list are explicitly listed as discrete chemicals [14, 23].

International frameworks have been put in place to foster chemical weapons nonproliferation and disarmament. These frameworks feature lists of chemicals that can be used as chemical weapons or precursors for their synthesis (CW-control lists). In these lists, chemicals of concern are described through chemical names and CAS Registry Numbers®. Importantly, in some CW-control lists, some entries, rather than specifying individual chemicals, describe families of related chemicals. Working with CW-control lists poses challenges for frontline customs and export control officers implementing these frameworks. Entries that describe families of chemicals are not easy to interpret, especially for non-chemists. Moreover, synonyms and chemical variants complicate the issue of checking CW-control lists through names and registry numbers. To ameliorate these problems, we have developed a functioning prototype of a cheminformatics tool that automates the task of assessing whether a chemical is pa...