SMILES to Smell: Decoding the Structure-Odor Relationship of Chemical Compounds Using the Deep Neural Network Approach

Predicting odor from vibrational spectra: a data-driven approach

This study investigates olfaction, a complex and not well-understood sensory modality. The chemical mechanism behind smell can be described by so far proposed two theories: vibrational and docking theories. The vibrational theory has been gaining acceptance lately but needs more extensive validation. To fill this gap for the first time, we, with the help of data-driven classification, clustering, and Explainable AI techniques, systematically analyze a large dataset of vibrational spectra (VS) of 3018 molecules obtained from the atomistic simulation. The study utlizes image representations of VS using Gramian Angular Fields and Markov Transition Fields, allowing computer vision techniques to be applied for better feature extraction and improved odor classification. Furthermore, we fuse the PCA-reduced fingerprint features with image features, which show additional improvement in classification results. We use two clustering methods, agglomerative hierarchical (AHC) and k-means, on dimensionality reduced (UMAP, MDS, t-SNE, and PCA) VS and image features, which shed further insight into the connections between molecular structure, VS, and odor. Additionally, we contrast our method with an earlier work that employed traditional machine learning on fingerprint features for the same dataset, and demonstrate that even with a representative subset of 3018 molecules, our deep learning model outperforms previous results. This comprehensive and systematic analysis highlights the potential of deep learning in furthering the field of olfactory research while confirming the vibrational theory of olfaction.

Artificial intelligence and cheminformatics tools: a contribution to the drug development and chemical science

In the ever-evolving field of drug discovery, the integration of Artificial Intelligence (AI) and Machine Learning (ML) with cheminformatics has proven to be a powerful combination. Cheminformatics, which combines the principles of computer science and chemistry, is used to extract chemical information and search compound databases, while the application of AI and ML allows for the identification of potential hit compounds, optimization of synthesis routes, and prediction of drug efficacy and toxicity. This collaborative approach has led to the discovery, preclinical evaluations and approval of over 70 drugs in recent years. To aid researchers in the pursuit of new drugs, this article presents a comprehensive list of databases, datasets, predictive and generative models, scoring functions and web platforms that have been launched between 2021 and 2022. These resources provide a wealth of information and tools for computer-assisted drug development, and are a valuable asset for those working in the field of cheminformatics. Overall, the integration of AI, ML and cheminformatics has greatly advanced the drug discovery process and continues to hold great potential for the future. As new resources and technologies become available, we can expect to see even more groundbreaking discoveries and advancements in these fields.Communicated by Ramaswamy H. Sarma.

A deep position-encoding model for predicting olfactory perception from molecular structures and electrostatics

Predicting olfactory perceptions from odorant molecules is challenging due to the complex and potentially discontinuous nature of the perceptual space for smells. In this study, we introduce a deep learning model, Mol-PECO (Molecular Representation by Positional Encoding of Coulomb Matrix), designed to predict olfactory perceptions based on molecular structures and electrostatics. Mol-PECO learns the efficient embedding of molecules by utilizing the Coulomb matrix, which encodes atomic coordinates and charges, as an alternative of the adjacency matrix and its Laplacian eigenfunctions as positional encoding of atoms. With a comprehensive dataset of odor molecules and descriptors, Mol-PECO outperforms traditional machine learning methods using molecular fingerprints and graph neural networks based on adjacency matrices. The learned embeddings by Mol-PECO effectively capture the odor space, enabling global clustering of descriptors and local retrieval of similar odorants. This work contributes to a deeper understanding of the olfactory sense and its mechanisms.

Predicting Odor Sensory Attributes of Unidentified Chemicals in Water Using Fragmentation Mass Spectra with Machine Learning Models

Knowing odor sensory attributes of odorants lies at the core of odor tracking when addressing waterborne odor issues. However, experimental determination covering tens of thousands of odorants in authentic water is not pragmatic due to the complexity of odorant identification and odor evaluation. In this study, we propose the first machine learning (ML) model to predict odor perception/threshold aiming at odorants in water, which can use either molecular structure or MS2 spectra as input features. We demonstrate that model performance using MS2 spectra is nearly as good as that using unequivocal structures, both with outstanding accuracy. We particularly show the model's robustness in predicting odor sensory attributes of unidentified chemicals by using the experimentally obtained MS2 spectra from nontarget analysis on authentic water samples. Interpreting the developed models, we identify the intricate interaction of functional groups as the predominant influence factor on odor sensory attributes. We also highlight the important roles of carbon chain length, molecular weight, etc., in the inherent olfactory mechanisms. These findings streamline the odor sensory attribute prediction and are crucial advancements toward credible tracking and efficient control of off-odors in water.

Regression Study of Odorant Chemical Space, Molecular Structural Diversity, and Natural Language Description

Odor is analyzed on the human olfactometry systems in various steps. The mapping from chemical structures to olfactory perceptions of smell is an extremely challenging task. Scientists have been unable to find a measure to distinguish the perceptual similarity between odorants. In this study, we report regression analysis and visualization based on the odorant chemical space. We discuss the relation between the odor descriptors and their structural diversity for odorants groups associated with each odor descriptor. We studied the influence of structural diversity on the odor descriptor predictability. The results suggest that the diversity of molecular structures, which is associated with the same odor descriptor, is related to the resolutional confusion with the odor descriptor.

Predicting Binding Energies and Electronic Properties of Boron Nitride Fullerenes Using a Graph Convolutional Network

As isoelectronic counterparts of carbon fullerenes, medium-sized boron nitride clusters also prefer cage structures composed of even-sized polygons. As the cluster size increases, the number of cage isomers grows rapidly, and determining the ground state structure requires a tremendous amount of DFT calculations. Herein, we develop a graph convolutional network (GCN) that can describe the energy of a (BN)n cage by its topology connection. We define a vertex feature vector on a dual polyhedron by the permutation of the neighbor vertices' degree and aggregate the information on vertices by two graph convolutional layers to learn the local feature of the dual polyhedron. The GCN is trained on (BN)28 and subsequently tested on (BN)23 and (BN)24 data sets, which satisfactorily reproduce the order of isomer energies from DFT calculations. We further employ the trained GCN to predict the ground state structures within the size range of n = 25-32, which agree well with DFT results. Using the same GCN framework, we also successfully trained the highest-occupied or lowest-unoccupied orbital energies of (BN)28 isomers. The present graph convolutional network establishes a direct mapping between the topological connection and the energetic or electronic properties of a cage-like cluster or molecule.

The neural representations of valence transformation in indole processing

Indole is often associated with a sweet and floral odor typical of jasmine flowers at low concentrations and an unpleasant, animal-like odor at high concentrations. However, the mechanism whereby the brain processes this opposite valence of indole is not fully understood yet. In this study, we aimed to investigate the neural mechanisms underlying indole valence encoding in conversion and nonconversion groups using the smelling task to arouse pleasantness. For this purpose, 12 conversion individuals and 15 nonconversion individuals participated in an event-related functional magnetic resonance imaging paradigm with low (low-indole) and high (high-indole) indole concentrations in which valence was manipulated independent of intensity. The results of this experiment showed that neural activity in the right amygdala, orbitofrontal cortex and insula was associated with valence independent of intensity. Furthermore, activation in the right orbitofrontal cortex in response to low-indole was positively associated with subjective pleasantness ratings. Conversely, activation in the right insula and amygdala in response to low-indole was positively correlated with anticipatory hedonic traits. Interestingly, while amygdala activation in response to high-indole also showed a positive correlation with these hedonic traits, such correlation was observed solely with right insula activation in response to high-indole. Additionally, activation in the right amygdala in response to low-indole was positively correlated with consummatory pleasure and hedonic traits. Regarding olfactory function, only activation in the right orbitofrontal cortex in response to high-indole was positively correlated with olfactory identification, whereas activation in the insula in response to low-indole was negatively correlated with the level of self-reported olfactory dysfunction. Based on these findings, valence transformation of indole processing in the right orbitofrontal cortex, insula, and amygdala may be associated with individual hedonic traits and perceptual differences.

Data science opportunities of large language models for neuroscience and biomedicine

Large language models (LLMs) are a new asset class in the machine-learning landscape. Here we offer a primer on defining properties of these modeling techniques. We then reflect on new modes of investigation in which LLMs can be used to reframe classic neuroscience questions to deliver fresh answers. We reason that LLMs have the potential to (1) enrich neuroscience datasets by adding valuable meta-information, such as advanced text sentiment, (2) summarize vast information sources to overcome divides between siloed neuroscience communities, (3) enable previously unthinkable fusion of disparate information sources relevant to the brain, (4) help deconvolve which cognitive concepts most usefully grasp phenomena in the brain, and much more.

Mixtures Recomposition by Neural Nets: A Multidisciplinary Overview

Artificial Neural Networks (ANNs) are transforming how we understand chemical mixtures, providing an expressive view of the chemical space and multiscale processes. Their hybridization with physical knowledge can bridge the gap between predictivity and understanding of the underlying processes. This overview explores recent progress in ANNs, particularly their potential in the 'recomposition' of chemical mixtures. Graph-based representations reveal patterns among mixture components, and deep learning models excel in capturing complexity and symmetries when compared to traditional Quantitative Structure-Property Relationship models. Key components, such as Hamiltonian networks and convolution operations, play a central role in representing multiscale mixtures. The integration of ANNs with Chemical Reaction Networks and Physics-Informed Neural Networks for inverse chemical kinetic problems is also examined. The combination of sensors with ANNs shows promise in optical and biomimetic applications. A common ground is identified in the context of statistical physics, where ANN-based methods iteratively adapt their models by blending their initial states with training data. The concept of mixture recomposition unveils a reciprocal inspiration between ANNs and reactive mixtures, highlighting learning behaviors influenced by the training environment.

Molecular dynamics simulation techniques and their application to aroma compounds/cyclodextrin inclusion complexes: A review

Homeostatic technologies play a crucial role in maintaining the quality and extending the service life of aroma compounds (ACs). Commercial cyclodextrins (CDs) are commonly used to form inclusion complexes (ICs) with ACs to enhance their solubility, stability, and morphology. The selection of suitable CDs and ACs is of paramount importance in this process. Molecular dynamics (MD) simulations provide an in-depth understanding of the interactions between ACs and CDs, aiding researchers in optimising the properties and effects of ICs. This review offers a systematic discussion of the application of MD simulations in ACs/CDs ICs, covering the establishment of the simulation process, parameter selection, model evaluation, and various application cases, along with their advantages and disadvantages. Additionally, this review summarises the major achievements and challenges of this method while identifying areas that require further exploration. These findings may contribute to a comprehensive understanding of the formation and stabilization mechanisms of ACs/CDs ICs and offer guidance for the selection and computational characterisation of CDs in the AC steady state.

Deep learning algorithms applied to computational chemistry

Recently, there has been a significant increase in the use of deep learning techniques in the molecular sciences, which have shown high performance on datasets and the ability to generalize across data. However, no model has achieved perfect performance in solving all problems, and the pros and cons of each approach remain unclear to those new to the field. Therefore, this paper aims to review deep learning algorithms that have been applied to solve molecular challenges in computational chemistry. We proposed a comprehensive categorization that encompasses two primary approaches; conventional deep learning and geometric deep learning models. This classification takes into account the distinct techniques employed by the algorithms within each approach. We present an up-to-date analysis of these algorithms, emphasizing their key features and open issues. This includes details of input descriptors, datasets used, open-source code availability, task solutions, and actual research applications, focusing on general applications rather than specific ones such as drug discovery. Furthermore, our report discusses trends and future directions in molecular algorithm design, including the input descriptors used for each deep learning model, GPU usage, training and forward processing time, model parameters, the most commonly used datasets, libraries, and optimization schemes. This information aids in identifying the most suitable algorithms for a given task. It also serves as a reference for the datasets and input data frequently used for each algorithm technique. In addition, it provides insights into the benefits and open issues of each technique, and supports the development of novel computational chemistry systems.

Developing human olfactory network and exploring olfactory receptor-odorant interaction

The Olfactory receptor (OR)-odorant interactions are perplexed. ORs can bind to structurally diverse odorants associated with one or more odor percepts. Various attempts have been made to understand the intricacies of OR-odorant interaction. In this study, experimentally documented OR-odorant interactions are investigated comprehensively to; (a) suggest potential odor percepts for ORs based on the OR-OR network; (b) determine how odorants interacting with specific ORs differ in terms of inherent pharmacophoric features and molecular properties, (c) identify molecular interactions that explained OR-odorant interactions of selective ORs; and (d) predict the probable role of ORs other than olfaction. Human olfactory receptor network (hORnet) is developed to study possible odor percepts for ORs. We identified six molecular properties which showed variation and significant patterns to differentiate odorants binding with five ORs. The pharmacophore analysis revealed that odorants subset of five ORs follow similar pharmacophore hypothesis, (one hydrogen acceptor and two hydrophobic regions) but differ in terms of distance and orientation of pharmacophoric features. To ascertain the binding site residues and key interactions between the selected ORs and their interacting odorants, 3D-structure modelling, docking and molecular dynamics studies were carried out. Lastly, the potential role of ORs beyond olfaction is explored. A human OR-OR network was developed to suggest possible odor percepts for ORs using empirically proven OR-odorant interactions. We sought to find out significant characteristics, molecular properties, and molecular interactions that could explain OR-odorant interactions and add to the understanding of the complex issue of odor perception.Communicated by Ramaswamy H. Sarma.

XGBoost odor prediction model: finding the structure-odor relationship of odorant molecules using the extreme gradient boosting algorithm

Determining the structure-odor relationship has always been a very challenging task. The main challenge in investigating the correlation between the molecular structure and its associated odor is the ambiguous and obscure nature of verbally defined odor descriptors, particularly when the odorant molecules are from different sources. With the recent developments in machine learning (ML) technology, ML and data analytic techniques are significantly being used for quantitative structure-activity relationship (QSAR) in the chemistry domain toward knowledge discovery where the traditional Edisonian methods have not been useful. The smell perception of odorant molecules is one of the aforementioned tasks, as olfaction is one of the least understood senses as compared to other senses. In this study, the XGBoost odor prediction model was generated to classify smells of odorant molecules from their SMILES strings. We first collected the dataset of 1278 odorant molecules with seven basic odor descriptors, and then 1875 physicochemical properties of odorant molecules were calculated. To obtain relevant physicochemical features, a feature reduction algorithm called PCA was also employed. The ML model developed in this study was able to predict all seven basic smells with high precision (>99%) and high sensitivity (>99%) when tested on an independent test dataset. The results of the proposed study were also compared with three recently conducted studies. The results indicate that the XGBoost-PCA model performed better than the other models for predicting common odor descriptors. The methodology and ML model developed in this study may be helpful in understanding the structure-odor relationship.Communicated by Ramaswamy H. Sarma.

Deep learning-based prediction model for diagnosing gastrointestinal diseases using endoscopy images

BACKGROUND

Gastrointestinal (GI) infections are quite common today around the world. Colonoscopy or wireless capsule endoscopy (WCE) are noninvasive methods for examining the whole GI tract for abnormalities. Nevertheless, it requires a great deal of time and effort for doctors to visualize a large number of images, and diagnosis is prone to human error. As a result, developing automated artificial intelligence (AI) based GI disease diagnosis methods is a crucial and emerging research area. AI-based prediction models may lead to improvements in the early diagnosis of gastrointestinal disorders, assessing severity, and healthcare systems for the benefit of patients as well as clinicians. The focus of this research is on the early diagnosis of gastrointestinal diseases using a convolution neural network (CNN) to enhance diagnosis accuracy.

METHODS

Various CNN models (baseline model and using transfer learning (VGG16, InceptionV3, and ResNet50)) were trained on a benchmark image dataset, KVASIR, containing images from inside the GI tract using n-fold cross-validation. The dataset comprises images of three disease states-polyps, ulcerative colitis, and esophagitis-as well as images of the healthy colon. Data augmentation strategies together with statistical measures were used to improve and evaluate the model's performance. Additionally, the test set comprising 1200 images was used to evaluate the model's accuracy and robustness.

RESULTS

The CNN model using the weights of the ResNet50 pre-trained model achieved the highest average accuracy of approximately 99.80% on the training set (100% precision and approximately 99% recall) and accuracies of 99.50% and 99.16% on the validation and additional test set, respectively, while diagnosing GI diseases. When compared to other existing systems, the proposed ResNet50 model outperforms them all.

CONCLUSION

The findings of this study indicate that AI-based prediction models using CNNs, specifically ResNet50, can improve diagnostic accuracy for detecting gastrointestinal polyps, ulcerative colitis, and esophagitis. The prediction model is available at https://github.com/anjus02/GI-disease-classification.git.

A critical review on odor measurement and prediction

Odor pollution has become a global environmental issue of increasing concern in recent years. Odor measurements are the basis of assessing and solving odor problems. Olfactory and chemical analysis can be used for odor and odorant measurements. Olfactory analysis reflects the subjective perception of human, and chemical analysis reveals the chemical composition of odors. As an alternative to olfactory analysis, odor prediction methods have been developed based on chemical and olfactory analysis results. The combination of olfactory and chemical analysis is the best way to control odor pollution, evaluate the performances of the technologies, and predict odor. However, there are still some limitations and obstacles for each method, their combination, and the prediction. Here, we present an overview of odor measurement and prediction. Different olfactory analysis methods (namely, the dynamic olfactometry method and the triangle odor bag method) are compared in detail, the latest revisions of the standard olfactometry methods are summarized, and the uncertainties of olfactory measurement results (i.e., the odor thresholds) are analyzed. The researches, applications, and limitations of chemical analysis and odor prediction are introduced and discussed. Finally, the development and application of odor databases and algorithms for optimizing odor measurement and prediction methods are prospected, and a preliminary framework for an odor database is proposed. This review is expected to provide insights into odor measurement and prediction.

Profiling and annotation of carbonyl compounds in Baijiu Daqu by chlorine isotope labeling-assisted ultrahigh-performance liquid chromatography-high resolution mass spectrometry

Carbonyl compounds are among the most important flavor substances that affect the taste of Baijiu. However, high coverage analysis of carbonyl compounds is obstructed due to the poor ionization efficiency of these compounds. Here we report a chlorine isotope labeling-assisted ultrahigh-performance liquid chromatography-high resolution mass spectrometry-based method (CIL-UHPLCHRMS) for profiling and annotation of carbonyl compounds in sauce flavored-Baijiu Daqu. 4-Chloro-2-hydrazinylpyridine was demonstrated to be a good labeling reagent that could achieve highly sensitive profiling and high-coverage screening of carbonyl compounds in the absence of heavy isotope labeling reagents. In the analysis of eight carbonyl standards representing different carbonyl categories, l-(-)-fucose, 2-carboxybenzaldehyde, 2-hydroxyacetophenone and heptan-2-one could be ionized only after labeling and MS signals were significantly increased for other 4 standards with an enhancement factor ranging from 181-fold for 3-methoxysalicylaldehyde to 3141-fold for tridecan-2-one. The annotation was achieved based on multidimensional information including MS1, predicted t_R, in silico MS/MS and manually annotated fragments. In total, 487 carbonyl compounds were detected in Baijiu Daqu, among which, 314 (64.5%) of them were positively or putatively identified. The outcome of the linearity (with a linear range of 2, 3 orders of magnitude), precision (less than 10%), and limit of detection (varied from 0.07 to 0.10 nM) indicated that the method was adequate for profiling carbonyl compounds in complex biological samples. The established method was successfully applied to study carbonyl compounds in Baijiu Daqu with different colors and different seasons. Taken collectively, the present work provides an effective, simple and economic strategy for comprehensive analysis of carbonyl compounds in complex matrix samples.

Artificial intelligence in intestinal polyp and colorectal cancer prediction

Artificial intelligence (AI) algorithms and their application to disease detection and decision support for healthcare professions have greatly evolved in the recent decade. AI has been widely applied and explored in gastroenterology for endoscopic analysis to diagnose intestinal cancers, premalignant polyps, gastrointestinal inflammatory lesions, and bleeding. Patients' responses to treatments and prognoses have both been predicted using AI by combining multiple algorithms. In this review, we explored the recent applications of AI algorithms in the identification and characterization of intestinal polyps and colorectal cancer predictions. AI-based prediction models have the potential to help medical practitioners diagnose, establish prognoses, and find accurate conclusions for the treatment of patients. With the understanding that rigorous validation of AI approaches using randomized controlled studies is solicited before widespread clinical use by health authorities, the article also discusses the limitations and challenges associated with deploying AI systems to diagnose intestinal malignancies and premalignant lesions.

Combining the Classification and Pharmacophore Approaches to Understand Homogeneous Olfactory Perceptions at Peripheral Level: Focus on Two Aroma Mixtures

The mechanisms involved in the homogeneous perception of odorant mixtures remain largely unknown. With the aim of enhancing knowledge about blending and masking mixture perceptions, we focused on structure-odor relationships by combining the classification and pharmacophore approaches. We built a dataset of about 5000 molecules and their related odors and reduced the multidimensional space defined by 1014 fingerprints representing the structures to a tridimensional 3D space using uniform manifold approximation and projection (UMAP). The self-organizing map (SOM) classification was then performed using the 3D coordinates in the UMAP space that defined specific clusters. We explored the allocating in these clusters of the components of two aroma mixtures: a blended mixture (red cordial (RC) mixture, 6 molecules) and a masking binary mixture (isoamyl acetate/whiskey-lactone [IA/WL]). Focusing on clusters containing the components of the mixtures, we looked at the odor notes carried by the molecules belonging to these clusters and also at their structural features by pharmacophore modeling (PHASE). The obtained pharmacophore models suggest that WL and IA could have a common binding site(s) at the peripheral level, but that would be excluded for the components of RC. In vitro experiments will soon be carried out to assess these hypotheses.

OWSum: algorithmic odor prediction and insight into structure-odor relationships

We derived and implemented a linear classification algorithm for the prediction of a molecule's odor, called Olfactory Weighted Sum (OWSum). Our approach relies solely on structural patterns of the molecules as features for algorithmic treatment and uses conditional probabilities combined with tf-idf values. In addition to the prediction of molecular odor, OWSum provides insights into properties of the dataset and allows to understand how algorithmic classifications are reached by quantitatively assigning structural patterns to odors. This provides chemists with an intuitive understanding of underlying interactions. To deal with ambiguities of the natural language used to describe odor, we introduced descriptor overlap as a metric for the quantification of semantic overlap between descriptors. Thus, grouping of descriptors and derivation of higher-level descriptors becomes possible. Our approach poses a large leap forward in our capabilities to understand and predict molecular features.

Smelling the Disease: Diagnostic Potential of Breath Analysis

Breath analysis is a relatively recent field of research with much promise in scientific and clinical studies. Breath contains endogenously produced volatile organic components (VOCs) resulting from metabolites of ingested precursors, gut and air-passage bacteria, environmental contacts, etc. Numerous recent studies have suggested changes in breath composition during the course of many diseases, and breath analysis may lead to the diagnosis of such diseases. Therefore, it is important to identify the disease-specific variations in the concentration of breath to diagnose the diseases. In this review, we explore methods that are used to detect VOCs in laboratory settings, VOC constituents in exhaled air and other body fluids (e.g., sweat, saliva, skin, urine, blood, fecal matter, vaginal secretions, etc.), VOC identification in various diseases, and recently developed electronic (E)-nose-based sensors to detect VOCs. Identifying such VOCs and applying them as disease-specific biomarkers to obtain accurate, reproducible, and fast disease diagnosis could serve as an alternative to traditional invasive diagnosis methods. However, the success of VOC-based identification of diseases is limited to laboratory settings. Large-scale clinical data are warranted for establishing the robustness of disease diagnosis. Also, to identify specific VOCs associated with illness states, extensive clinical trials must be performed using both analytical instruments and electronic noses equipped with stable and precise sensors.

Application of artificial intelligence to decode the relationships between smell, olfactory receptors and small molecules

Deciphering the relationship between molecules, olfactory receptors (ORs) and corresponding odors remains a challenging task. It requires a comprehensive identification of ORs responding to a given odorant. With the recent advances in artificial intelligence and the growing research in decoding the human olfactory perception from chemical features of odorant molecules, the applications of advanced machine learning have been revived. In this study, Convolutional Neural Network (CNN) and Graphical Convolutional Network (GCN) models have been developed on odorant molecules-odors and odorant molecules-olfactory receptors using a large set of 5955 molecules, 160 odors and 106 olfactory receptors. The performance of such models is promising with a Precision/Recall Area Under Curve of 0.66 for the odorant-odor and 0.91 for the odorant-olfactory receptor GCN models respectively. Furthermore, based on the correspondence of odors and ORs associated for a set of 389 compounds, an odor-olfactory receptor pairwise score was computed for each odor-OR combination allowing to suggest a combinatorial relationship between olfactory receptors and odors. Overall, this analysis demonstrate that artificial intelligence may pave the way in the identification of the smell perception and the full repertoire of receptors for a given odorant molecule.

Predicting odor from molecular structure: a multi-label classification approach

Decoding the factors behind odor perception has long been a challenge in the field of human neuroscience, olfactory research, perfumery, psychology, biology and chemistry. The new wave of data-driven and machine learning approaches to predicting molecular properties are a growing area of research interest and provide for significant improvement over conventional statistical methods. We look at these approaches in the context of predicting molecular odor, specifically focusing on multi-label classification strategies employed for the same. Namely binary relevance, classifier chains, and random forests adapted to deal with such a task. This challenge, termed quantitative structure–odor relationship, remains an unsolved task in the field of sensory perception in machine learning, and we hope to emulate the results achieved in the field of vision and auditory perception in olfaction over time.

Insight into the Structure-Odor Relationship of Molecules: A Computational Study Based on Deep Learning

Molecules with pleasant odors, unacceptable odors, and even serious toxicity are closely related to human social life. It is impractical to identify the odors of molecules in large quantities (particularly hazardous odors) using experimental methods. Computer-aided methods have currently attracted increasing attention for the prediction of molecular odors. Here, through models based on multilayer perceptron (MLP) and physicochemical descriptors (MLP-Des), MLP and molecular fingerprint, and convolutional neural network (CNN), we conduct the two-class prediction of odor/no odor, fruity/no odor, floral/no odor, and woody/no odor, and the multi-class prediction of fruity/flowery/woody/no odor on our newly refined molecular odor datasets. We show that three kinds of predictors can robustly predict molecular odors. The MLP-Des model not only exhibits the best prediction results (the AUC values are 0.99 and 0.86 for the two- and multi-classification models, respectively) but can also well reflect the characteristics of the structure–odor relationship of molecules. The CNN model takes 2D molecular images as input and can automatically extract the structural features related to molecular odors. The proposed models are of great help for the prediction of molecular odorants, understanding the underlying relationship between chemical structure and odor perception, and the discovery of new odorous and/or hazardous molecules.

Impact of the Interactions between Fragrances and Cosmetic Bases on the Fragrance Olfactory Performance: A Tentative to Correlate SPME-GC/MS Analysis with That of an Experienced Perfumer

“Seta e Ciliegia” and “Narguilé” fragrances were mixed to form a binary blend with chemically stable, non-volatile, odourless, simple bases of different lipophilicity widely used in skin care and hair care formulations, such as caprylic-capric triglyceride, glycerine, paraffin, dimethicone, isopropyl myristate and butylene glycol, with the objective to verify how the olfactory performance of fragrances can be influenced by skin or hair care ingredients. The semiquantitative approach applied in this study aims in providing a practical solution to appropriately combine a fragrance with cosmetic ingredients. Pure fragrance and binary blends were analysed by solid phase microextraction gas chromatography tandem mass spectrometry (SPME-GC/MS), based on the assumption that the solid phase microextraction is able to extract volatile compounds, mimicking the ability of the nose to capture similar volatile compounds. Fifty-seven and forty-four compounds were identified by SPME-GC/MS in pure fragrances “Seta e Ciliegia” and “Narguilé”, respectively. Once mixed with the bases, the analysis of the blends revealed that a qualitative modification in the chromatograms could occur according to the characteristics of the bases. In general, for both fragrances, blends with glycerin and butylene glycol, which are the most hydrophilic bases among the ones tested, were able to release most of the peaks, that were thus still present in the chromatograms. Differently, in the blends with caprylic-capric triglyceride, most of the peaks are lost. Blends with paraffine, dimethicone and isopropyl myristate showed an intermediate behaviour. These results were thus compared with the sensory evaluation made by an experienced perfumer, capable of assessing the different olfactory performances of pure fragrances and their different binary blends. The evaluation made by the perfumer fitted well with the analytical results, and in the blends where most of the peaks were revealed in the chromatogram, the perfumer found a similar olfactory profile for example with glycerin, butylene glycol, while a modification of the olfactory profile was highlighted when several peaks were not still present in the chromatogram, as it was the case with caprylic-capric triglyceride. Interestingly, when the most typical peaks of a fragrance were still observed in the blend, even if some of them were lost, the olfactory performance was not lost, as was the case of paraffin and isopropyl myristate. In the case of dimethicone, its high volatility was considered responsible for a certain decrease in the fragrance “volume”. The results achieved with this investigation can be used to hypothesize that the different compounds of a fragrance, characterized for the first time by different volatility and solubility, could be differently retained by the bases: the more lipophilic are strongly retained by the lipophilic bases with a consequently reduced volatility that limits the possibility of being appreciated by the nose and that corresponds to disappearance or a percentage reduction from the chromatogram. Therefore, in a more accurate and helpful view for a formulator, we could come to the conclusion that based on the results achieved by our investigation, the inclusion of a less lipophilic base can be more appropriate to exalt more lipophilic fragrances.

Synthesis of Cyclic Fragrances via Transformations of Alkenes, Alkynes and Enynes: Strategies and Recent Progress

With increasing demand for customized commodities and the greater insight and understanding of olfaction, the synthesis of fragrances with diverse structures and odor characters has become a core task. Recent progress in organic synthesis and catalysis enables the rapid construction of carbocycles and heterocycles from readily available unsaturated molecular building blocks, with increased selectivity, atom economy, sustainability and product diversity. In this review, synthetic methods for creating cyclic fragrances, including both natural and synthetic ones, will be discussed, with a focus on the key transformations of alkenes, alkynes, dienes and enynes. Several strategies will be discussed, including cycloaddition, catalytic cyclization, ring-closing metathesis, intramolecular addition, and rearrangement reactions. Representative examples and the featured olfactory investigations will be highlighted, along with some perspectives on future developments in this area.

DeePred-BBB: A Blood Brain Barrier Permeability Prediction Model With Improved Accuracy

The blood-brain barrier (BBB) is a selective and semipermeable boundary that maintains homeostasis inside the central nervous system (CNS). The BBB permeability of compounds is an important consideration during CNS-acting drug development and is difficult to formulate in a succinct manner. Clinical experiments are the most accurate method of measuring BBB permeability. However, they are time taking and labor-intensive. Therefore, numerous efforts have been made to predict the BBB permeability of compounds using computational methods. However, the accuracy of BBB permeability prediction models has always been an issue. To improve the accuracy of the BBB permeability prediction, we applied deep learning and machine learning algorithms to a dataset of 3,605 diverse compounds. Each compound was encoded with 1,917 features containing 1,444 physicochemical (1D and 2D) properties, 166 molecular access system fingerprints (MACCS), and 307 substructure fingerprints. The prediction performance metrics of the developed models were compared and analyzed. The prediction accuracy of the deep neural network (DNN), one-dimensional convolutional neural network, and convolutional neural network by transfer learning was found to be 98.07, 97.44, and 97.61%, respectively. The best performing DNN-based model was selected for the development of the “DeePred-BBB” model, which can predict the BBB permeability of compounds using their simplified molecular input line entry system (SMILES) notations. It could be useful in the screening of compounds based on their BBB permeability at the preliminary stages of drug development. The DeePred-BBB is made available at https://github.com/12rajnish/DeePred-BBB.

Recent Applications of Artificial Intelligence in Early Cancer Detection

Cancer is a deadly disease often caused by the accumulation of various genetic mutations and pathological alterations. The death rate can only be reduced when it is detected in the early stages because treatment of cancer when the tumor has not metastasized in many regions of the body is more effective. However, early cancer detection is fraught with difficulties. Advances in artificial intelligence (AI) have developed a new scope for efficient and early detection of such a fatal disease. AI algorithms have a remarkable ability to perform well on a variety of tasks that are presented or fed to the system. Numerous studies have produced machine learning and deep learning-assisted cancer prediction models to detect cancer from previously accessible data with better accuracy, sensitivity, and specificity. It has been observed that the accuracy of prediction models in classifying fed data as benign, malignant, or normal is improved by implementing efficient image processing techniques and data segmentation augmentation methodologies, along with advanced algorithms. In this review, recent AI-based models for the diagnosis of the most prevalent cancers in the breast, lung, brain, and skin have been analysed. Available AI techniques, data preparation, modeling processes, and performance assessments have been included in the review.

OlfactionBase: a repository to explore odors, odorants, olfactory receptors and odorant-receptor interactions

Olfaction is a multi-stage process that initiates with the odorants entering the nose and terminates with the brain recognizing the odor associated with the odorant. In a very intricate way, the process incorporates various components functioning together and in synchronization. OlfactionBase is a free, open-access web server that aims to bring together knowledge about many aspects of the olfaction mechanism in one place. OlfactionBase contains detailed information of components like odors, odorants, and odorless compounds with physicochemical and ADMET properties, olfactory receptors (ORs), odorant- and pheromone binding proteins, OR-odorant interactions in Human and Mus musculus. The dynamic, user-friendly interface of the resource facilitates exploration of different entities: finding chemical compounds having desired odor, finding odorants associated with OR, associating chemical features with odor and OR, finding sequence information of ORs and related proteins. Finally, the data in OlfactionBase on odors, odorants, olfactory receptors, human and mouse OR-odorant pairs, and other associated proteins could aid in the inference and improved understanding of odor perception, which might provide new insights into the mechanism underlying olfaction. The OlfactionBase is available at https://bioserver.iiita.ac.in/olfactionbase/.

Recent trends in stem cell-based therapies and applications of artificial intelligence in regenerative medicine

Stem cells are undifferentiated cells that can self-renew and differentiate into diverse types of mature and functional cells while maintaining their original identity. This profound potential of stem cells has been thoroughly investigated for its significance in regenerative medicine and has laid the foundation for cell-based therapies. Regenerative medicine is rapidly progressing in healthcare with the prospect of repair and restoration of specific organs or tissue injuries or chronic disease conditions where the body’s regenerative process is not sufficient to heal. In this review, the recent advances in stem cell-based therapies in regenerative medicine are discussed, emphasizing mesenchymal stem cell-based therapies as these cells have been extensively studied for clinical use. Recent applications of artificial intelligence algorithms in stem cell-based therapies, their limitation, and future prospects are highlighted.

Bitter Taste and Olfactory Receptors: Beyond Chemical Sensing in the Tongue and the Nose

Abstract

The Up-and-Coming-Scientist section of the current issue of the Journal of Membrane Biology features the invited essay by Dr. Mercedes Alfonso-Prieto, Assistant Professor at the Forschungszentrum Jülich (FZJ), Germany, and the Heinrich-Heine University Düsseldorf, Vogt Institute for Brain Research. Dr. Alfonso-Prieto completed her doctoral degree in chemistry at the Barcelona Science Park, Spain, in 2009, pursued post-doctoral research in computational molecular sciences at Temple University, USA, and then, as a Marie Curie post-doctoral fellow at the University of Barcelona, worked on computations of enzyme reactions and modeling of photoswitchable ligands targeting neuronal receptors. In 2016, she joined the Institute for Advanced Science and the Institute for Computational Biomedicine at the FZJ, where she pursues research on modeling and simulation of chemical senses. The invited essay by Dr. Alfonso-Prieto discusses state-of-the-art modeling of molecular receptors involved in chemical sensing – the senses of taste and smell. These receptors, and computational methods to study them, are the focus of Dr. Alfonso-Prieto’s research. Recently, Dr. Alfonso-Prieto and colleagues have presented a new methodology to predict ligand binding poses for GPCRs, and extensive computations that deciphered the ligand selectivity determinants of bitter taste receptors. These developments inform our current understanding of how taste occurs at the molecular level.

Graphic Abstract

Progress on open chemoinformatic tools for expanding and exploring the chemical space

The concept of chemical space is a cornerstone in chemoinformatics, and it has broad conceptual and practical applicability in many areas of chemistry, including drug design and discovery. One of the most considerable impacts is in the study of structure-property relationships where the property can be a biological activity or any other characteristic of interest to a particular chemistry discipline. The chemical space is highly dependent on the molecular representation that is also a cornerstone concept in computational chemistry. Herein, we discuss the recent progress on chemoinformatic tools developed to expand and characterize the chemical space of compound data sets using different types of molecular representations, generate visual representations of such spaces, and explore structure-property relationships in the context of chemical spaces. We emphasize the development of methods and freely available tools focusing on drug discovery applications. We also comment on the general advantages and shortcomings of using freely available and easy-to-use tools and discuss the value of using such open resources for research, education, and scientific dissemination.

De novo Design of G Protein-Coupled Receptor 40 Peptide Agonists for Type 2 Diabetes Mellitus Based on Artificial Intelligence and Site-Directed Mutagenesis

G protein-coupled receptor 40 (GPR40), one of the G protein-coupled receptors that are available to sense glucose metabolism, is an attractive target for the treatment of type 2 diabetes mellitus (T2DM). Despite many efforts having been made to discover small-molecule agonists, there is limited research focus on developing peptides acting as GPR40 agonists to treat T2DM. Here, we propose a novel strategy for peptide design to generate and determine potential peptide agonists against GPR40 efficiently. A molecular fingerprint similarity (MFS) model combined with a deep neural network (DNN) and convolutional neural network was applied to predict the activity of peptides constructed by unnatural amino acids (UAAs). Site-directed mutagenesis (SDM) further optimized the peptides to form specific favorable interactions, and subsequent flexible docking showed the details of the binding mechanism between peptides and GPR40. Molecular dynamics (MD) simulations further verified the stability of the peptide–protein complex. The R-square of the machine learning model on the training set and the test set reached 0.87 and 0.75, respectively; and the three candidate peptides showed excellent performance. The strategy based on machine learning and SDM successfully searched for an optimal design with desirable activity comparable with the model agonist in phase III clinical trials.

Vibration-based biomimetic odor classification

Olfaction is not as well-understood as vision or audition, nor technologically addressed. Here, Chemical Graph Theory is shown to connect the vibrational spectrum of an odorant molecule, invoked in the Vibration Theory of Olfaction, to its structure, which is germane to the orthodox Shape Theory. Atomistic simulations yield the Eigen-VAlue (EVA) vibrational pseudo-spectra for 20 odorant molecules grouped into 6 different ‘perceptual’ classes by odour. The EVA is decomposed into peaks corresponding to different types of vibrational modes. A novel secondary pseudo-spectrum, informed by this physical insight—the Peak-Decomposed EVA (PD-EVA)—has been proposed here. Unsupervised Machine Learning (spectral clustering), applied to the PD-EVA, clusters the odours into different ‘physical’ (vibrational) classes that match the ‘perceptual’, and also reveal inherent perceptual subclasses. This establishes a physical basis for vibration-based odour classification, harmonizes the Shape and Vibration theories, and points to vibration-based sensing as a promising path towards a biomimetic electronic nose.

Remodelling structure-based drug design using machine learning

To keep up with the pace of rapid discoveries in biomedicine, a plethora of research endeavors had been directed toward Rational Drug Development that slowly gave way to Structure-Based Drug Design (SBDD). In the past few decades, SBDD played a stupendous role in identification of novel drug-like molecules that are capable of altering the structures and/or functions of the target macromolecules involved in different disease pathways and networks. Unfortunately, post-delivery drug failures due to adverse drug interactions have constrained the use of SBDD in biomedical applications. However, recent technological advancements, along with parallel surge in clinical research have led to the concomitant establishment of other powerful computational techniques such as Artificial Intelligence (AI) and Machine Learning (ML). These leading-edge tools with the ability to successfully predict side-effects of a wide range of drugs have eventually taken over the field of drug design. ML, a subset of AI, is a robust computational tool that is capable of data analysis and analytical model building with minimal human intervention. It is based on powerful algorithms that use huge sets of 'training data' as inputs to predict new output values, which improve iteratively through experience. In this review, along with a brief discussion on the evolution of the drug discovery process, we have focused on the methodologies pertaining to the technological advancements of machine learning. This review, with specific examples, also emphasises the tremendous contributions of ML in the field of biomedicine, while exploring possibilities for future developments.

Olfactory Perception in Relation to the Physicochemical Odor Space

A growing body of research aims at solving what is often referred to as the stimulus-percept problem in olfactory perception. Although computational efforts have made it possible to predict perceptual impressions from the physicochemical space of odors, studies with large psychophysical datasets from non-experts remain scarce. Following previous approaches, we developed a physicochemical odor space using 4094 molecular descriptors of 1389 odor molecules. For 20 of these odors, we examined associations with perceived pleasantness, intensity, odor quality and detection threshold, obtained from a dataset of 2000 naïve participants. Our results show significant differences in perceptual ratings, and we were able to replicate previous findings on the association between perceptual ratings and the first dimensions of the physicochemical odor space. However, the present analyses also revealed striking interindividual variations in perceived pleasantness and intensity. Additionally, interactions between pleasantness, intensity, and olfactory and trigeminal qualitative dimensions were found. To conclude, our results support previous findings on the relation between structure and perception on the group level in our sample of non-expert raters. In the challenging task to relate olfactory stimulus and percept, the physicochemical odor space can serve as a reliable and helpful tool to structure the high-dimensional space of olfactory stimuli. Nevertheless, human olfactory perception in the individual is not an analytic process of molecule detection alone, but is part of a holistic integration of multisensory inputs, context and experience.

A deep neural network-based approach for prediction of mutagenicity of compounds

We are exposed to various chemical compounds present in the environment, cosmetics, and drugs almost every day. Mutagenicity is a valuable property that plays a significant role in establishing a chemical compound's safety. Exposure and handling of mutagenic chemicals in the environment pose a high health risk; therefore, identification and screening of these chemicals are essential. Considering the time constraints and the pressure to avoid laboratory animals' use, the shift to alternative methodologies that can establish a rapid and cost-effective detection without undue over-conservation seems critical. In this regard, computational detection and identification of the mutagens in environmental samples like drugs, pesticides, dyes, reagents, wastewater, cosmetics, and other substances is vital. From the last two decades, there have been numerous efforts to develop the prediction models for mutagenicity, and by far, machine learning methods have demonstrated some noteworthy performance and reliability. However, the accuracy of such prediction models has always been one of the major concerns for the researchers working in this area. The mutagenicity prediction models were developed using deep neural network (DNN), support vector machine, k-nearest neighbor, and random forest. The developed classifiers were based on 3039 compounds and validated on 1014 compounds; each of them encoded with 1597 molecular feature vectors. DNN-based prediction model yielded highest prediction accuracy of 92.95% and 83.81% with the training and test data, respectively. The area under the receiver's operating curve and precision-recall curve values were found to be 0.894 and 0.838, respectively. The DNN-based classifier not only fits the data with better performance as compared to traditional machine learning algorithms, viz., support vector machine, k-nearest neighbor, and random forest (with and without feature reduction) but also yields better performance metrics. In current work, we propose a DNN-based model to predict mutagenicity of compounds.