Machine-Learning-Based Olfactometer: Prediction of Odor Perception from Physicochemical Features of Odorant Molecules

Fit-for-purpose WWTP unmanned aerial systems: A game changer towards an integrated and sustainable management strategy

In the ongoing Anthropocene era, air quality monitoring constitutes a primary axis of European and international policies for all sectors, including Waste Water Treatment Plants (WWTPs). Unmanned Aerial Systems (UASs) with proper sensing equipment provide an edge technology for air quality and odor monitoring. In addition, Unmanned Aerial Vehicle (UAV) photogrammetry has been used in civil engineering, environmental (water) quality assessment and lately for industrial facilities monitoring. This study constitutes a systematic review of the late advances and limitations of germane equipment and implementations. Despite their unassailable flexibility and efficiency, the employment of the aforementioned technologies in WWTP remote monitoring is yet sparse, partial, and concerns only particular aspects. The main finding of the review was the lack of a tailored UAS for WWTP monitoring in the literature. Therefore, to fill in this gap, we propose a fit-for-purpose remote monitoring system consisting of a UAS with a platform that would integrate all the required sensors for air quality (i.e., emissions of H2S, NH3, NOx, SO2, CH4, CO, CO2, VOCs, and PM) and odor monitoring, multispectral and thermal cameras for photogrammetric structural health monitoring (SHM) and wastewater/effluent properties (e.g., color, temperature, etc.) of a WWTP. It constitutes a novel, supreme and integrated approach to improve the sustainable management of WWTPs. Specifically, the developments that a fit-for-purpose WWTP UAS would launch, are fostering the decision-making of managers, administrations, and policymakers, both in operational conditions and in case of failures, accidents or natural disasters. Furthermore, it would significantly reduce the operational expenditure of a WWTP, ensuring personnel and population health standards, and local area sustainability.

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.

Prediction on Air-Nasal Mucus Partition Coefficients of Odor Compounds

The air-nasal mucus partition coefficient is a crucial property among all of the interaction mechanisms between odor molecules and olfactory receptors, since this property contributes to our sense of smell. Due to the complexity of the mucus composition, in vivo determination of the air-mucus partition coefficient is a technical challenge. A predictable model of the air-mucus partition coefficient can provide valuable insights into the chemical properties that govern olfactory perception and can help design desired odorants. In this study, we propose a novel model based on the deep-layer neural network (DNN) algorithm to predict the air-mucus partition coefficients for a range of odor compounds. The molecular surface charge density (σ-profile) calculated from the COnductor like Screening MOdel for Real Solvents (COSMO-RS) thermodynamic package was adapted as descriptors of structural features of odor molecules. The results revealed that the air-mucus partition coefficients are highly correlated to the σ-profile of the studied compounds. The information obtained from the study provided interpretable results, which not only help in identifying the molecular features that contribute to the air-mucus partition coefficient of odorants but also aid in the design of compounds with the desired odor properties.

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.

Predicting odor profile of food from its chemical composition: Towards an approach based on artificial intelligence and flavorists expertise

Odor is central to food quality. Still, a major challenge is to understand how the odorants present in a given food contribute to its specific odor profile, and how to predict this olfactory outcome from the chemical composition. In this proof-of-concept study, we seek to develop an integrative model that combines expert knowledge, fuzzy logic, and machine learning to predict the quantitative odor description of complex mixtures of odorants. The model output is the intensity of relevant odor sensory attributes calculated on the basis of the content in odor-active comounds. The core of the model is the mathematically formalized knowledge of four senior flavorists, which provided a set of optimized rules describing the sensory-relevant combinations of odor qualities the experts have in mind to elaborate the target odor sensory attributes. The model first queries analytical and sensory databases in order to standardize, homogenize, and quantitatively code the odor descriptors of the odorants. Then the standardized odor descriptors are translated into a limited number of odor qualities used by the experts thanks to an ontology. A third step consists of aggregating all the information in terms of odor qualities across all the odorants found in a given product. The final step is a set of knowledge-based fuzzy membership functions representing the flavorist expertise and ensuring the prediction of the intensity of the target odor sensory descriptors on the basis of the products' aggregated odor qualities; several methods of optimization of the fuzzy membership functions have been tested. Finally, the model was applied to predict the odor profile of 16 red wines from two grape varieties for which the content in odorants was available. The results showed that the model can predict the perceptual outcome of food odor with a certain level of accuracy, and may also provide insights into combinations of odorants not mentioned by the experts.

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.

Comparing molecular representations, e-nose signals, and other featurization, for learning to smell aroma molecules

Recent research has attempted to predict our perception of odorants using Machine Learning models. The featurization of the olfactory stimuli usually represents the odorants using molecular structure parameters, molecular fingerprints, mass spectra, or e-nose signals. However, the impact of the choice of featurization on predictive performance remains poorly reported in direct comparative studies. This paper experiments with different sensory features for several olfactory perception tasks. We investigate the multilabel classification of aroma molecules in odor descriptors. We investigate single-label classification not only in fine-grained odor descriptors ('orange', 'waxy', etc.), but also in odor descriptor groups. We created a database of odor vectors for 114 aroma molecules to conduct our experiments using a QCM (Quartz Crystal Microbalance) type smell sensor module (Aroma Coder®V2 Set). We compare these smell features with different baseline features to evaluate the cluster composition, considering the frequencies of the top odor descriptors carried by the aroma molecules. Experimental results suggest a statistically significant better performance of the QCM type smell sensor module compared with other baseline features with F1 evaluation metric.

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.

Investigation of the structure-odor relationship using a Transformer model

The relationships between molecular structures and their properties are subtle and complex, and the properties of odor are no exception. Molecules with similar structures, such as a molecule and its optical isomer, may have completely different odors, whereas molecules with completely distinct structures may have similar odors. Many works have attempted to explain the molecular structure-odor relationship from chemical and data-driven perspectives. The Transformer model is widely used in natural language processing and computer vision, and the attention mechanism included in the Transformer model can identify relationships between inputs and outputs. In this paper, we describe the construction of a Transformer model for predicting molecular properties and interpreting the prediction results. The SMILES data of 100,000 molecules are collected and used to predict the existence of molecular substructures, and our proposed model achieves an F1 value of 0.98. The attention matrix is visualized to investigate the substructure annotation performance of the attention mechanism, and we find that certain atoms in the target substructures are accurately annotated. Finally, we collect 4462 molecules and their odor descriptors and use the proposed model to infer 98 odor descriptors, obtaining an average F1 value of 0.33. For the 19 odor descriptors that achieved F1 values greater than 0.45, we also attempt to summarize the relationship between the molecular substructures and odor quality through the attention matrix.

Unraveling the Thermal Oxidation Products and Peroxidation Mechanisms of Different Chemical Structures of Lipids: An Example of Molecules Containing Oleic Acid

Lipid structures affect lipid oxidation, causing differences in types and contents of volatiles and nonvolatiles in various foods. In this study, the oxidation differences of monoacylglycerol (MAG), triacylglycerol (TAG), phosphatidylethanolamine (PE), and phosphatidylcholine (PC) with oleoyl residues and oleic acid (FA) during thermal treatment were investigated. Volatiles and nonvolatiles were monitored by gas chromatography-mass spectrometry and ultrahigh-performance liquid chromatography-Q-Exactive HF-X Orbitrap Mass Spectrometer, respectively. The results showed that the structures of MAG and TAG could delay the chain initiation reaction. The polar heads of PC and PE remarkably influenced the oxidation rate and the formation of the oxidation products probably due to the hydrogen bonds formed with free radicals. Among the volatile oxidation products, aldehydes, acids, and furans with eight or nine carbon atoms accounted for the majority in FA, MAG, TAG, and PC samples, but PE samples mainly generated ketones with nine or 10 carbon atoms. The formation of nonvolatile products in TAG samples possessed significant stage-specific changes. Fatty acid esters of hydroxy fatty acids were only produced in the free fatty acid oxidation model. The activity of chemical bonds participating in the truncation reaction decreased to both sides from the double bond position.

Exploration of sensing data to realize intended odor impression using mass spectrum of odor mixture

Recently, olfactory information on odorants has been associated with their corresponding molecular features. Such information has been obtained by predicting the sensory test evaluation scores from the molecular structure parameters or the sensing data. On the other hand, we develop a method of the prediction of molecular features corresponding to the odor impression. We utilize a machine-learning-based odor predictive model introduced in our previous research, and we propose a mathematical model for exploring the sensing data space. By using mass spectrum as sensing data in the predictive model, we can represent predicted mass spectrum as those of an odor mixture, and the mixing ratio can be obtained. We show that the mass spectrum of apple flavor with enhanced 'fruit' and 'sweet' impressions can be obtained using 59 and 60 molecules respectively by using our analysis method.

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.

Imputation of sensory properties using deep learning

Predicting the sensory properties of compounds is challenging due to the subjective nature of the experimental measurements. This testing relies on a panel of human participants and is therefore also expensive and time-consuming. We describe the application of a state-of-the-art deep learning method, Alchemite™, to the imputation of sparse physicochemical and sensory data and compare the results with conventional quantitative structure-activity relationship methods and a multi-target graph convolutional neural network. The imputation model achieved a substantially higher accuracy of prediction, with improvements in R² between 0.26 and 0.45 over the next best method for each sensory property. We also demonstrate that robust uncertainty estimates generated by the imputation model enable the most accurate predictions to be identified and that imputation also more accurately predicts activity cliffs, where small changes in compound structure result in large changes in sensory properties. In combination, these results demonstrate that the use of imputation, based on data from less expensive, early experiments, enables better selection of compounds for more costly studies, saving experimental time and resources.

Identification of the key target profiles underlying the drugs of narrow therapeutic index for treating cancer and cardiovascular disease

An appropriate therapeutic index is crucial for drug discovery and development since narrow therapeutic index (NTI) drugs with slight dosage variation may induce severe adverse drug reactions or potential treatment failure. To date, the shared characteristics underlying the targets of NTI drugs have been explored by several studies, which have been applied to identify potential drug targets. However, the association between the drug therapeutic index and the related disease has not been dissected, which is important for revealing the NTI drug mechanism and optimizing drug design. Therefore, in this study, two classes of disease (cancers and cardiovascular disorders) with the largest number of NTI drugs were selected, and the target property of the corresponding NTI drugs was analyzed. By calculating the biological system profiles and human protein–protein interaction (PPI) network properties of drug targets and adopting an AI-based algorithm, differentiated features between two diseases were discovered to reveal the distinct underlying mechanisms of NTI drugs in different diseases. Consequently, ten shared features and four unique features were identified for both diseases to distinguish NTI from NNTI drug targets. These computational discoveries, as well as the newly found features, suggest that in the clinical study of avoiding narrow therapeutic index in those diseases, the ability of target to be a hub and the efficiency of target signaling in the human PPI network should be considered, and it could thus provide novel guidance in the drug discovery and clinical research process and help to estimate the drug safety of cancer and cardiovascular disease.

Data based predictive models for odor perception

Machine learning and data analytics are being increasingly used for quantitative structure property relation (QSPR) applications in the chemical domain where the traditional Edisonian approach towards knowledge-discovery have not been fruitful. The perception of odorant stimuli is one such application as olfaction is the least understood among all the other senses. In this study, we employ machine learning based algorithms and data analytics to address the efficacy of using a data-driven approach to predict the perceptual attributes of an odorant namely the odorant characters (OC) of "sweet" and "musky". We first analyze a psychophysical dataset containing perceptual ratings of 55 subjects to reveal patterns in the ratings given by subjects. We then use the data to train several machine learning algorithms such as random forest, gradient boosting and support vector machine for prediction of the odor characters and report the structural features correlating well with the odor characters based on the optimal model. Furthermore, we analyze the impact of the data quality on the performance of the models by comparing the semantic descriptors generally associated with a given odorant to its perception by majority of the subjects. The study presents a methodology for developing models for odor perception and provides insights on the perception of odorants by untrained human subjects and the effect of the inherent bias in the perception data on the model performance. The models and methodology developed here could be used for predicting odor characters of new odorants.

Prediction of Odor Concentration Emitted from Wastewater Treatment Plant Using an Artificial Neural Network (ANN)

The odor emitted from a wastewater treatment plant (WWTP) is an important environmental problem. An estimation of odor emission rate is difficult to detect and quantify. To address this, various approaches including the development of emission factors and measurement using a closed chamber have been employed. However, the evaluation of odor emission involves huge manpower, time, and cost. An artificial neural network (ANN) is recognized as an efficient method to find correlations between nonlinear data and prediction of future data based on these correlations. Due to its usefulness, ANN is used to solve complicated problems in various disciplines of sciences and engineering. In this study, a method to predict the odor concentration in a WWTP using ANN was developed. The odor concentration emitted from a WWTP was predicted by the ANN based on water quality data such as biological oxygen demand, dissolved oxygen, and pH. The water quality and odor concentration data from the WWTP were measured seasonally in spring, summer, and autumn and these were used as input variations to the ANN model. The odor predicted by the ANN model was compared with the measured data and the prediction accuracy was estimated. Suggestions for improving prediction accuracy are presented.

Predicting human odor perception represented by continuous values from mass spectra of essential oils resembling chemical mixtures

There have been recent advances in predicting odor characteristics using molecular structure parameters of chemicals. Although the molecular structure parameters are available for each chemical, they cannot be used for chemical mixtures. This study will elucidate a computational method of predicting human odor perception from the mass spectra of chemical mixtures such as essential oils. Furthermore, a method for obtaining similarity among odor descriptors has been proposed although the dataset contains binary values only. When the database indicates a set of odor descriptors for one sample, only binary data are available and the correlation between the similar descriptors disappears. Thus, the prediction performance degrades for not considering the similarity among the odor descriptors. Since mass spectra dataset is highly dimensional, we use auto-encoder to learn the compressed representation from the mass spectra of essential oils in its bottleneck hidden layer and then accomplishes the hierarchical clustering to create odor descriptor groups with similar odor impressions using a matrix of continuous value-based correlation coefficient as well as natural language processing. This work will help to expatiate the process of overcoming binary value problem and find out the similarity among odor descriptors using machine learning with natural language semantic representation of words. To overcome the problem of disproportionate ratio of positive and negative class for both the continuous value-based correlation coefficient and word similarity based models, we use Synthetic Minority Oversampling Technique (SMOTE). This model allows us to predict human odor perception through computer simulations by forming odor descriptors group. Accordingly, this study demonstrates the feasibility of ensembling machine learning with natural language processing and SMOTE approach for predicting odor descriptor group from mass spectra of essential oils.

Chemical features mining provides new descriptive structure-odor relationships

An important goal in researching the biology of olfaction is to link the perception of smells to the chemistry of odorants. In other words, why do some odorants smell like fruits and others like flowers? While the so-called stimulus-percept issue was resolved in the field of color vision some time ago, the relationship between the chemistry and psycho-biology of odors remains unclear up to the present day. Although a series of investigations have demonstrated that this relationship exists, the descriptive and explicative aspects of the proposed models that are currently in use require greater sophistication. One reason for this is that the algorithms of current models do not consistently consider the possibility that multiple chemical rules can describe a single quality despite the fact that this is the case in reality, whereby two very different molecules can evoke a similar odor. Moreover, the available datasets are often large and heterogeneous, thus rendering the generation of multiple rules without any use of a computational approach overly complex. We considered these two issues in the present paper. First, we built a new database containing 1689 odorants characterized by physicochemical properties and olfactory qualities. Second, we developed a computational method based on a subgroup discovery algorithm that discriminated perceptual qualities of smells on the basis of physicochemical properties. Third, we ran a series of experiments on 74 distinct olfactory qualities and showed that the generation and validation of rules linking chemistry to odor perception was possible. Taken together, our findings provide significant new insights into the relationship between stimulus and percept in olfaction. In addition, by automatically extracting new knowledge linking chemistry of odorants and psychology of smells, our results provide a new computational framework of analysis enabling scientists in the field to test original hypotheses using descriptive or predictive modeling.

An important issue in olfaction sciences deals with the question of how a chemical information can be translated into percepts. This is known as the stimulus-percept problem. Here, we set out to better understand this issue by combining knowledge about the chemistry and cognition of smells with computational olfaction. We also assumed that not only one, but several physicochemical models may describe a given olfactory quality. To achieve this aim, a first challenge was to set up a database with ~1700 molecules characterized by chemical features and described by olfactory qualities (e.g. fruity, woody). A second challenge consisted in developing a computational model enabling the discrimination of olfactory qualities based on these chemical features. By meeting these 2 challenges, we provided for several olfactory qualities new chemical models describing why an odorant molecule smells fruity or woody (among others). For most qualities, multiple (rather than a single) chemical models were generated. These findings provide new elements of knowledge about the relationship between odorant chemistry and perception. They also make it possible to envisage concrete applications in the aroma and fragrance field where chemical characterization of smells is an important step in the design of new products.

Machine Learning in Human Olfactory Research

The complexity of the human sense of smell is increasingly reflected in complex and high-dimensional data, which opens opportunities for data-driven approaches that complement hypothesis-driven research. Contemporary developments in computational and data science, with its currently most popular implementation as machine learning, facilitate complex data-driven research approaches. The use of machine learning in human olfactory research included major approaches comprising 1) the study of the physiology of pattern-based odor detection and recognition processes, 2) pattern recognition in olfactory phenotypes, 3) the development of complex disease biomarkers including olfactory features, 4) odor prediction from physico-chemical properties of volatile molecules, and 5) knowledge discovery in publicly available big databases. A limited set of unsupervised and supervised machine-learned methods has been used in these projects, however, the increasing use of contemporary methods of computational science is reflected in a growing number of reports employing machine learning for human olfactory research. This review provides key concepts of machine learning and summarizes current applications on human olfactory data.

Determining the Balance Between Drug Efficacy and Safety by the Network and Biological System Profile of Its Therapeutic Target

One of the most challenging puzzles in drug discovery is the identification and characterization of candidate drug of well-balanced profile between efficacy and safety. So far, extensive efforts have been made to evaluate this balance by estimating the quantitative structure–therapeutic relationship and exploring target profile of adverse drug reaction. Particularly, the therapeutic index (TI) has emerged as a key indicator illustrating this delicate balance, and a clinically successful agent requires a sufficient TI suitable for it corresponding indication. However, the TI information are largely unknown for most drugs, and the mechanism underlying the drugs with narrow TI (NTI drugs) is still elusive. In this study, the collective effects of human protein–protein interaction (PPI) network and biological system profile on the drugs' efficacy–safety balance were systematically evaluated. First, a comprehensive literature review of the FDA approved drugs confirmed their NTI status. Second, a popular feature selection algorithm based on artificial intelligence (AI) was adopted to identify key factors differencing the target mechanism between NTI and non-NTI drugs. Finally, this work revealed that the targets of NTI drugs were highly centralized and connected in human PPI network, and the number of similarity proteins and affiliated signaling pathways of the corresponding targets was much higher than those of non-NTI drugs. These findings together with the newly discovered features or feature groups clarified the key factors indicating drug's narrow TI, and could thus provide a novel direction for determining the delicate drug efficacy-safety balance.