UMPred-FRL: A New Approach for Accurate Prediction of Umami Peptides Using Feature Representation Learning

Unraveling novel umami peptides from yeast extract (Saccharomyces cerevisiae) using peptidomics and molecular interaction modeling

Yeast extract is increasingly becoming an attractive source for unraveling novel umami peptides that are healthier and more nutritious than traditional seasonings. In the present study, a strategy for screening novel umami peptides was established using mass spectrometry-based peptidomics combined with molecular interaction modeling, emphasizing on smaller peptides than previously reported. Four representative novel umami peptides of FE, YDQ, FQEY, and SPFSQ from yeast extract (Saccharomyces cerevisiae) were identified and validated by sensory evaluation, with thresholds determined as 0.234 ± 0.045, 0.576 ± 0.175, 0.327 ± 0.057 and 0.456 ± 0.070 mmol/L, respectively. Hydrogen and ionic bonds were the main characteristic interactions between the umami peptides and the well-recognized receptor T1R1/T1R3, in which Asp 110, Thr 112, Arg 114, Arg 240, Lys 342, and Glu 264 were the key sites in ligand-receptor recognition. Our study provides accurate sequences of umami peptides and molecular interaction mechanism for the umami effect.

Advancements in production, assessment, and food applications of salty and saltiness-enhancing peptides: A review

Salt is important for food flavor, but excessive sodium intake leads to adverse health consequences. Thus, salty and saltiness-enhancing peptides are developed for sodium-reduction products. This review elucidates saltiness perception process and analyses correlation between the peptide structure and saltiness-enhancing ability. These peptides interact with taste receptors to produce saltiness perception, including ENaC, TRPV1, and TMC4. This review also outlines preparation, isolation, purification, characterization, screening, and assessment techniques of these peptides and discusses their potential applications. These peptides are from various sources and produced through enzymatic hydrolysis, microbial fermentation, or Millard reaction and then separated, purified, identified, and screened. Sensory evaluation, electronic tongue, bioelectronic tongue, and cell and animal models are the primary saltiness assessment approaches. These peptides can be used in sodium-reduction food products to produce "clean label" items, and the peptides with biological activity can also serve as functional ingredients, making them very promising for food industry.

VmmScore: An umami peptide prediction and receptor matching program based on a deep learning approach

Peptides, with recognized physiological and medical implications, such as the ability to lower blood pressure and lipid levels, are central to our research on umami taste perception. This study introduces a computational strategy to tackle the challenge of identifying optimal umami receptors for these peptides. Our VmmScore algorithm includes two integral components: Mlp4Umami, a predictive module that evaluates the umami taste potential of peptides, and mm-Score, which enhances the receptor matching process through a machine learning-optimized molecular docking and scoring system. This system encompasses the optimization of docking structures, clustering of umami peptides, and a comparative analysis of docking energies across peptide clusters, streamlining the receptor identification process. Employing machine learning, our method offers a strategic approach to the intricate task of umami receptor determination. We undertook virtual screening of peptides derived from Lateolabrax japonicus, experimentally verifying the umami taste of three identified peptides and determining their corresponding receptors. This work not only advances our understanding of the mechanisms behind umami taste perception but also provides a rapid and cost-effective method for peptide screening. The source code is publicly accessible at https://github.com/heyigacu/mlp4umami/, encouraging further scientific exploration and collaborative efforts within the research community.

Characteristics of saltiness-enhancing peptides derived from yeast proteins and elucidation of their mechanism of action by molecular docking

This study aimed to identify saltiness-enhancing peptides from yeast protein and elucidate their mechanisms by molecular docking. Yeast protein hydrolysates with optimal saltiness-enhancing effects were prepared under conditions determined using an orthogonal test. Ten saltiness-enhancing peptide candidates were screened using an integrated virtual screening strategy. Sensory evaluation demonstrated that these peptides exhibited diverse taste characteristics (detection thresholds: 0.13-0.50 mmol/L). Peptides NKF, LGLR, WDL, NMKF, FDSL and FDGK synergistically or additively enhanced the saltiness of a 0.30% NaCl solution. Molecular docking revealed that these peptides predominantly interacted with TMC4 by hydrogen bonding, with hydrophilic amino acids from both peptides and TMC4 playing a pivotal role in their binding. Furthermore, Leu217, Gln377, Glu378, Pro474 and Cys475 were postulated as the key binding sites of TMC4. These findings establish a robust theoretical foundation for salt reduction strategies in food and provide novel insights into the potential applications of yeast proteins.

UmamiPreDL: Deep learning model for umami taste prediction of peptides using BERT and CNN

Taste is crucial in driving food choice and preference. Umami is one of the basic tastes defined by characteristic deliciousness and mouthfulness that it imparts to foods. Identification of ingredients to enhance umami taste is of significant value to food industry. Various models have been shown to predict umami taste using feature encodings derived from traditional molecular descriptors such as amphiphilic pseudo-amino acid composition, dipeptide composition, and composition-transition-distribution. Highest reported accuracy of 90.5 % was recently achieved through novel model architecture. Here, we propose use of biological sequence transformers such as ProtBert and ESM2, trained on the Uniref databases, as the feature encoders block. With combination of 2 encoders and 2 classifiers, 4 model architectures were developed. Among the 4 models, ProtBert-CNN model outperformed other models with accuracy of 95 % on 5-fold cross validation data and 94 % on independent data.

Predicting multiple taste sensations with a multiobjective machine learning method

Taste perception plays a pivotal role in guiding nutrient intake and aiding in the avoidance of potentially harmful substances through five basic tastes - sweet, bitter, umami, salty, and sour. Taste perception originates from molecular interactions in the oral cavity between taste receptors and chemical tastants. Hence, the recognition of taste receptors and the subsequent perception of taste heavily rely on the physicochemical properties of food ingredients. In recent years, several advances have been made towards the development of machine learning-based algorithms to classify chemical compounds' tastes using their molecular structures. Despite the great efforts, there remains significant room for improvement in developing multi-class models to predict the entire spectrum of basic tastes. Here, we present a multi-class predictor aimed at distinguishing bitter, sweet, and umami, from other taste sensations. The development of a multi-class taste predictor paves the way for a comprehensive understanding of the chemical attributes associated with each fundamental taste. It also opens the potential for integration into the evolving realm of multi-sensory perception, which encompasses visual, tactile, and olfactory sensations to holistically characterize flavour perception. This concept holds promise for introducing innovative methodologies in the rational design of foods, including pre-determining specific tastes and engineering complementary diets to augment traditional pharmacological treatments.

Exploring the Relationship between Small Peptides and the T1R1/T1R3 Umami Taste Receptor for Umami Peptide Prediction: A Combined Approach

Umami peptides are known for enhancing the taste experience by binding to oral umami T1R1 and T1R3 receptors. Among them, small peptides (composed of 2-4 amino acids) constitute nearly 40% of reported umami peptides. Given the diversity in amino acids and peptide sequences, umami small peptides possess tremendous untapped potential. By investigating 168,400 small peptides, we screened candidates binding to T1R1/T1R3 through molecular docking and molecular dynamics simulations, explored bonding types, amino acid characteristics, preferred binding sites, etc. Utilizing three-dimensional molecular descriptors, bonding information, and a back-propagation neural network, we developed a predictive model with 90.3% accuracy, identifying 24,539 potential umami peptides. Clustering revealed three classes with distinct logP (-2.66 ± 1.02, -3.52 ± 0.93, -2.44 ± 1.23) and asphericity (0.28 ± 0.12, 0.26 ± 0.11, 0.25 ± 0.11), indicating significant differences in shape and hydrophobicity (P < 0.05) among potential umami peptides binding to T1R1/T1R3. Following clustering, nine representative peptides (CQ, DP, NN, CSQ, DMC, TGS, DATE, HANR, and STAN) were synthesized and confirmed to possess umami taste through sensory evaluations and electronic tongue analyses. In summary, this study provides insights into exploring small peptide interactions with umami receptors, advancing umami peptide prediction models.

Bioinformatics and bioactive peptides from foods: Do they work together?

We live in the Big Data Era which affects many aspects of science, including research on bioactive peptides derived from foods, which during the last few decades have been a focus of interest for scientists. These two issues, i.e., the development of computer technologies and progress in the discovery of novel peptides with health-beneficial properties, are closely interrelated. This Chapter presents the example applications of bioinformatics for studying biopeptides, focusing on main aspects of peptide analysis as the starting point, including: (i) the role of peptide databases; (ii) aspects of bioactivity prediction; (iii) simulation of peptide release from proteins. Bioinformatics can also be used for predicting other features of peptides, including ADMET, QSAR, structure, and taste. To answer the question asked "bioinformatics and bioactive peptides from foods: do they work together?", currently it is almost impossible to find examples of peptide research with no bioinformatics involved. However, theoretical predictions are not equivalent to experimental work and always require critical scrutiny. The aspects of compatibility of in silico and in vitro results are also summarized herein.

Virtual screening and characteristics of novel umami peptides from porcine type I collagen

This study aimed to rapidly and precisely discover novel umami peptides from porcine type I collagen using virtual screening, sensory evaluation and molecular docking simulation. Porcine type I collagen was hydrolyzed in silico and six umami peptide candidates (CN, SM, CRD, GESMTDGF, MS, DGC) were shortlisted via umami taste, bioactivity, toxicity, allergenicity, solubility and stability predictions. The sensory evaluation confirmed that these peptides exhibited umami taste, with CRD, GESMTDGF and DGC displaying higher umami intensity and significant umami-enhancing effects in 0.35% sodium glutamate solution. Molecular docking predicted that Ser 276/384/385 of T1R1 and Asn68, Val277, Thr305, Ser306, Leu385 of T1R3 may also play critical roles in binding umami peptides. The umami taste of peptides may be perceived mainly through the formation of hydrogen bonds with the hydrophilic amino acids of T1R1/T1R3. This work provided a robust procedure and guidance to develop novel umami peptides from food byproducts.

Invited review: Camel milk-derived bioactive peptides and diabetes-Molecular view and perspectives

In dairy science, camel milk (CM) constitutes a center of interest for scientists due to its known beneficial effect on diabetes as demonstrated in many in vitro, in vivo, and clinical studies and trials. Overall, CM had positive effects on various parameters related to glucose transport and metabolism as well as the structural and functional properties of the pancreatic β-cells and insulin secretion. Thus, CM consumption may help manage diabetes; however, such a recommendation will become rationale and clinically conceivable only if the exact molecular mechanisms and pathways involved at the cellular levels are well understood. Moreover, the application of CM as an alternative antidiabetic tool may first require the identification of the exact bioactive molecules behind such antidiabetic properties. In this review, we describe the advances in our knowledge of the molecular mechanisms reported to be involved in the beneficial effects of CM in managing diabetes using different in vitro and in vivo models. This mainly includes the effects of CM on the different molecular pathways controlling (1) insulin receptor signaling and glucose uptake, (2) the pancreatic β-cell structure and function, and (3) the activity of key metabolic enzymes in glucose metabolism. Moreover, we described the current status of the identification of CM-derived bioactive peptides and their structure-activity relationship study and characterization in the context of molecular markers related to diabetes. Such an overview will not only enrich our scientific knowledge of the plausible mode of action of CM in diabetes but should ultimately rationalize the claim of the potential application of CM against diabetes. This will pave the way toward new directions and ideas for developing a new generation of antidiabetic products taking benefits from the chemical composition of CM.

Combining molecular docking and molecular dynamics simulation to discover four novel umami peptides from tuna skeletal myosin with sensory evaluation validation

Umami peptides are an important component of food flavoring agents and have high nutritional value. This work aimed to identify umami peptides from tuna skeletal myosin using a new model method of computer simulation, explore their umami mechanism, and further validate the umami tastes with sensory evaluation. Umami peptides LADW, MEIDD, VAEQE, and EEAEGT were discovered, and all of them bound to taste type 1 receptor 1 and receptor 3 via hydrogen bonds and van der Waals forces to form stable complexes. LADW exhibited the best affinity energy and binding capability. Sensory evaluation and electronic tongue confirmed that all peptides possessed an umami taste, and LADW exhibited the strongest umami intensity. This study not only explored four novel umami peptides to improve the value of tuna skeletal myosin but also provided a new method for the rapid discovery of umami peptides.

Identification of novel umami peptides from yeast extract and the mechanism against T1R1/T1R3

Yeast extract was separated by using ultrafiltration, gel filtration chromatography, and preparative high-performance liquid chromatography for analyzing the umami mechanism. 13 kinds of umami peptides were screened out from 73 kinds of peptides which were identified in yeast extract using nanoscale ultra-performance liquid chromatography-tandem mass spectrometry and virtual screening. The umami peptides were found to have a threshold range of 0.07-0.61 mM. DWTDDVEAR exhibited a strong umami taste with a pronounced enhancement effect for monosodium glutamate. Molecular docking studies revealed that specific amino acid residues in the T1R1 subunit, including Arg316, Ser401, and Asp315, played a critical role in the umami perception with these peptides. Overall, the study highlights the potential of natural flavor enhancers and provides insights into the mechanism of umami taste perception.

Classification of tastants: A deep learning based approach

Predicting the taste of molecules is of critical importance in the food and beverages, flavor, and pharmaceutical industries for the design and screening of new tastants. In this work, we have built deep learning models to classify sweet, bitter, and umami molecules- the three basic tastes whose sensation is mediated by G protein-coupled receptors. An extensive dataset containing 1466 bitter, 1764 sweet, and 238 umami tastants was curated from existing literature. We analyzed the chemical characteristics of the molecules, with special focus on the presence of different functional groups. A deep neural network model based on molecular descriptors and a graph neural network model were trained for taste prediction. The class imbalance due to fewer umami molecules was tackled using special sampling techniques. Both models show comparable performance during evaluation, but the graph-based model can learn task-specific representations from the molecular structure without requiring handcrafted features. We further explain the deep neural network predictions using Shapley additive explanations. Finally, we demonstrated the applicability of the models by screening bitter, sweet, and umami molecules from a large food database. This study develops an in-silico approach to classify molecules based on their taste by leveraging the recent progress in deep learning, which can serve as a powerful tool for tastant design.

Targeted deep learning classification and feature extraction for clinical diagnosis

Protein biomarkers can be used to characterize symptom classes, which describe the metabolic or immunodeficient state of patients during the progression of a specific disease. Recent literature has shown that machine learning methods can complement traditional clinical methods in identifying biomarkers. However, many machine learning frameworks only apply narrowly to a specific archetype or subset of diseases. In this paper, we propose a feature extractor which can discover protein biomarkers for a wide variety of classification problems. The feature extractor uses a special type of deep learning model, which discovers a latent space that allows for optimal class separation and enhanced class cluster identity. The extracted biomarkers can then be used to train highly accurate supervised learning models. We apply our methods to a dataset involving COVID-19 patients and another involving scleroderma patients, to demonstrate improved class separation and reduced false discovery rates compared to results obtained using traditional models.

Umami-BERT: An interpretable BERT-based model for umami peptides prediction

Umami peptides have received extensive attention due to their ability to enhance flavors and provide nutritional benefits. The increasing demand for novel umami peptides and the vast number of peptides present in food call for more efficient methods to screen umami peptides, and further exploration is necessary. Therefore, the purpose of this study is to develop deep learning (DL) model to realize rapid screening of umami peptides. The Umami-BERT model was devised utilizing a novel two-stage training strategy with Bidirectional Encoder Representations from Transformers (BERT) and the inception network. In the pre-training stage, attention mechanisms were implemented on a large amount of bioactive peptides sequences to acquire high-dimensional generalized features. In the re-training stage, umami peptide prediction was carried out on UMP789 dataset, which is developed through the latest research. The model achieved the performance with an accuracy (ACC) of 93.23% and MCC of 0.78 on the balanced dataset, as well as an ACC of 95.00% and MCC of 0.85 on the unbalanced dataset. The results demonstrated that Umami-BERT could predict umami peptides directly from their amino acid sequences and exceeded the performance of other models. Furthermore, Umami-BERT enabled the analysis of attention pattern learned by Umami-BERT model. The amino acids Alanine (A), Cysteine (C), Aspartate (D), and Glutamicacid (E) were found to be the most significant contributors to umami peptides. Additionally, the patterns of summarized umami peptides involving A, C, D, and E were analyzed based on the learned attention weights. Consequently, Umami-BERT exhibited great potential in the large-scale screening of candidate peptides and offers novel insight for the further exploration of umami peptides.

TIPred: a novel stacked ensemble approach for the accelerated discovery of tyrosinase inhibitory peptides

BACKGROUND

Tyrosinase is an enzyme involved in melanin production in the skin. Several hyperpigmentation disorders involve the overproduction of melanin and instability of tyrosinase activity resulting in darker, discolored patches on the skin. Therefore, discovering tyrosinase inhibitory peptides (TIPs) is of great significance for basic research and clinical treatments. However, the identification of TIPs using experimental methods is generally cost-ineffective and time-consuming.

RESULTS

Herein, a stacked ensemble learning approach, called TIPred, is proposed for the accurate and quick identification of TIPs by using sequence information. TIPred explored a comprehensive set of various baseline models derived from well-known machine learning (ML) algorithms and heterogeneous feature encoding schemes from multiple perspectives, such as chemical structure properties, physicochemical properties, and composition information. Subsequently, 130 baseline models were trained and optimized to create new probabilistic features. Finally, the feature selection approach was utilized to determine the optimal feature vector for developing TIPred. Both tenfold cross-validation and independent test methods were employed to assess the predictive capability of TIPred by using the stacking strategy. Experimental results showed that TIPred significantly outperformed the state-of-the-art method in terms of the independent test, with an accuracy of 0.923, MCC of 0.757 and an AUC of 0.977.

CONCLUSIONS

The proposed TIPred approach could be a valuable tool for rapidly discovering novel TIPs and effectively identifying potential TIP candidates for follow-up experimental validation. Moreover, an online webserver of TIPred is publicly available at http://pmlabstack.pythonanywhere.com/TIPred .

Sensory-Guided Isolation, Identification, and Active Site Calculation of Novel Umami Peptides from Ethanol Precipitation Fractions of Fermented Grain Wine (Huangjiu)

Huangjiu is rich in low-molecular-weight peptides and has an umami taste. In order for its umami peptides to be discovered, huangjiu was subjected to ultrafiltration, ethanol precipitation, and macroporous resin purification processes. The target fractions were gathered according to sensory evaluation. Subsequently, we used peptidomics to identify the sum of 4158 peptides in most umami fractions. Finally, six novel umami peptides (DTYNPR, TYNPR, SYNPR, RFRQGD, NFHHGD, and FHHGD) and five umami-enhancing peptides (TYNPR, SYNPR, NFHHGD, FHHGD, and TVDGPSH) were filtered via virtual screening, molecular docking, and sensory verification. Moreover, the structure-activity relationship was discussed using computational approaches. Docking analysis showed that all umami peptides tend to bind with T1R1 through hydrogen bonds and hydrophobic forces, which involve key residues HIS71, ASP147, ARG151, TYR220, SER276, and ALA302. The active site calculation revealed that the positions of the key umami residues D and R in the terminal may cause taste differences in identified peptides.

Classification-based machine learning approaches to predict the taste of molecules: A review

The capacity to discriminate safe from dangerous compounds has played an important role in the evolution of species, including human beings. Highly evolved senses such as taste receptors allow humans to navigate and survive in the environment through information that arrives to the brain through electrical pulses. Specifically, taste receptors provide multiple bits of information about the substances that are introduced orally. These substances could be pleasant or not according to the taste responses that they trigger. Tastes have been classified into basic (sweet, bitter, umami, sour and salty) or non-basic (astringent, chilling, cooling, heating, pungent), while some compounds are considered as multitastes, taste modifiers or tasteless. Classification-based machine learning approaches are useful tools to develop predictive mathematical relationships in such a way as to predict the taste class of new molecules based on their chemical structure. This work reviews the history of multicriteria quantitative structure-taste relationship modelling, starting from the first ligand-based (LB) classifier proposed in 1980 by Lemont B. Kier and concluding with the most recent studies published in 2022.

TROLLOPE: A novel sequence-based stacked approach for the accelerated discovery of linear T-cell epitopes of hepatitis C virus

Hepatitis C virus (HCV) infection is a concerning health issue that causes chronic liver diseases. Despite many successful therapeutic outcomes, no effective HCV vaccines are currently available. Focusing on T cell activity, the primary effector for HCV clearance, T cell epitopes of HCV (TCE-HCV) are considered promising elements to accelerate HCV vaccine efficacy. Thus, accurate and rapid identification of TCE-HCVs is recommended to obtain more efficient therapy for chronic HCV infection. In this study, a novel sequence-based stacked approach, termed TROLLOPE, is proposed to accurately identify TCE-HCVs from sequence information. Specifically, we employed 12 different sequence-based feature descriptors from heterogeneous perspectives, such as physicochemical properties, composition-transition-distribution information and composition information. These descriptors were used in cooperation with 12 popular machine learning (ML) algorithms to create 144 base-classifiers. To maximize the utility of these base-classifiers, we used a feature selection strategy to determine a collection of potential base-classifiers and integrated them to develop the meta-classifier. Comprehensive experiments based on both cross-validation and independent tests demonstrated the superior predictive performance of TROLLOPE compared with conventional ML classifiers, with cross-validation and independent test accuracies of 0.745 and 0.747, respectively. Finally, a user-friendly online web server of TROLLOPE (http://pmlabqsar.pythonanywhere.com/TROLLOPE) has been developed to serve research efforts in the large-scale identification of potential TCE-HCVs for follow-up experimental verification.

StackTTCA: a stacking ensemble learning-based framework for accurate and high-throughput identification of tumor T cell antigens

BACKGROUND

The identification of tumor T cell antigens (TTCAs) is crucial for providing insights into their functional mechanisms and utilizing their potential in anticancer vaccines development. In this context, TTCAs are highly promising. Meanwhile, experimental technologies for discovering and characterizing new TTCAs are expensive and time-consuming. Although many machine learning (ML)-based models have been proposed for identifying new TTCAs, there is still a need to develop a robust model that can achieve higher rates of accuracy and precision.

RESULTS

In this study, we propose a new stacking ensemble learning-based framework, termed StackTTCA, for accurate and large-scale identification of TTCAs. Firstly, we constructed 156 different baseline models by using 12 different feature encoding schemes and 13 popular ML algorithms. Secondly, these baseline models were trained and employed to create a new probabilistic feature vector. Finally, the optimal probabilistic feature vector was determined based the feature selection strategy and then used for the construction of our stacked model. Comparative benchmarking experiments indicated that StackTTCA clearly outperformed several ML classifiers and the existing methods in terms of the independent test, with an accuracy of 0.932 and Matthew's correlation coefficient of 0.866.

CONCLUSIONS

In summary, the proposed stacking ensemble learning-based framework of StackTTCA could help to precisely and rapidly identify true TTCAs for follow-up experimental verification. In addition, we developed an online web server ( http://2pmlab.camt.cmu.ac.th/StackTTCA ) to maximize user convenience for high-throughput screening of novel TTCAs.

Rapid screening based on machine learning and molecular docking of umami peptides from porcine bone

BACKGROUND

The traditional screening method for umami peptide, extracted from porcine bone, was labor-intensive and time-consuming. In this study, the rapid screening method and molecular mechanism of umami peptide was investigated.

RESULTS

This article showed that a more precisely rapid screening method with composite machine learning and molecular docking was used to screen the potential umami peptide from porcine bone. As reference, 24 reported umami peptides were predicated by composite machine learning, with the accuracy of 86.7%. In this study, potential umami peptide sequences from porcine bone were screened by UMPred-FRL, Umami-MRNN Demo, and molecular docking was used to provide further screening. Finally, nine peptides were screened and verified as umami peptides by this method: LREY, HEAL, LAKVH, FQKVVA, HVKELE, AEVKKAP, EAVEKPQS, KALSEEL and KKMFETES. The hydrogen bonding was deemed to be the main interaction force with receptor T1R3, and domain binding sites were Ser146, His121 and Glu277. The result demonstrated the feasibility of machine learning assisted T1R1/T1R3 receptor for rapid screening umami peptides. The screening method would not only adapt to screen umami peptides from porcine bone but possibly applied for other sources. It also provided a reference for rapid screening of umami peptides.

CONCLUSION

The manuscript lays a rapid screening method in screening umami peptide, and nine umami peptides from porcine bone were screened and identified. © 2022 Society of Chemical Industry.

Using a stacked ensemble learning framework to predict modulators of protein-protein interactions

Identifying small molecule protein-protein interaction modulators (PPIMs) is a highly promising and meaningful research direction for drug discovery, cancer treatment, and other fields. In this study, we developed a stacking ensemble computational framework, SELPPI, based on a genetic algorithm and tree-based machine learning method for effectively predicting new modulators targeting protein-protein interactions. More specifically, extremely randomized trees (ExtraTrees), adaptive boosting (AdaBoost), random forest (RF), cascade forest, light gradient boosting machine (LightGBM), and extreme gradient boosting (XGBoost) were used as basic learners. Seven types of chemical descriptors were taken as the input characteristic parameters. Primary predictions were obtained with each basic learner-descriptor pair. Then, the 6 methods mentioned above were used as meta learners and trained on the primary prediction in turn. The most efficient method was utilized as the meta learner. Finally, the genetic algorithm was used to select the optimal primary prediction output as the input of the meta learner for secondary prediction to obtain the final result. We systematically evaluated our model on the pdCSM-PPI datasets. To our knowledge, our model outperformed all existing models, which demonstrates its great power.

Conserved Sites and Recognition Mechanisms of T1R1 and T2R14 Receptors Revealed by Ensemble Docking and Molecular Descriptors and Fingerprints Combined with Machine Learning

Taste peptides, as an important component of protein-rich foodstuffs, potentiate the nutrition and taste of food. Thereinto, umami- and bitter-taste peptides have been ex tensively reported, while their taste mechanisms remain unclear. Meanwhile, the identification of taste peptides is still a time-consuming and costly task. In this study, 489 peptides with umami/bitter taste from TPDB (http://tastepeptides-meta.com/) were collected and used to train the classification models based on docking analysis, molecular descriptors (MDs), and molecular fingerprints (FPs). A consensus model, taste peptide docking machine (TPDM), was generated based on five learning algorithms (linear regression, random forest, gaussian naive bayes, gradient boosting tree, and stochastic gradient descent) and four molecular representation schemes. Model interpretive analysis showed that MDs (VSA_EState, MinEstateIndex, MolLogP) and FPs (598, 322, 952) had the greatest impact on the umami/bitter prediction of peptides. Based on the consensus docking results, we obtained the key recognition modes of umami/bitter receptors (T1Rs/T2Rs): (1) residues 107S-109S, 148S-154T, 247F-249A mainly form hydrogen bonding contacts and (2) residues 153A-158L, 163L, 181Q, 218D, 247F-249A in T1R1 and 56D, 106P, 107V, 152V-156F, 173K-180F in T2R14 constituted their hydrogen bond pockets. The model is available at http://www.tastepeptides-meta.com/yyds.

A Machine Learning Method to Identify Umami Peptide Sequences by Using Multiplicative LSTM Embedded Features

Umami peptides enhance the umami taste of food and have good food processing properties, nutritional value, and numerous potential applications. Wet testing for the identification of umami peptides is a time-consuming and expensive process. Here, we report the iUmami-DRLF that uses a logistic regression (LR) method solely based on the deep learning pre-trained neural network feature extraction method, unified representation (UniRep based on multiplicative LSTM), for feature extraction from the peptide sequences. The findings demonstrate that deep learning representation learning significantly enhanced the capability of models in identifying umami peptides and predictive precision solely based on peptide sequence information. The newly validated taste sequences were also used to test the iUmami-DRLF and other predictors, and the result indicates that the iUmami-DRLF has better robustness and accuracy and remains valid at higher probability thresholds. The iUmami-DRLF method can aid further studies on enhancing the umami flavor of food for satisfying the need for an umami-flavored diet.

Microbial succession and its effect on the formation of umami peptides during sufu fermentation

Sufu, a traditional Chinese fermented food, is famous for its unique flavor, especially umami. However, the formation mechanism of its umami peptides is still unclear. Here, we investigated the dynamic change of both umami peptides and microbial communities during sufu production. Based on peptidomic analysis, 9081 key differential peptides were identified, which mainly involved in amino acid transport and metabolism, peptidase activity and hydrolase activity. Twenty-six high-quality umami peptides with ascending trend were recognized by machine learning methods and Fuzzy c-means clustering. Then, through correlation analysis, five bacterial species (Enterococcus italicus, Leuconostoc citreum, L. mesenteroides, L. pseudomesenteroides, Tetragenococcus halophilus) and two fungi species (Cladosporium colombiae, Hannaella oryzae) were identified to be the core functional microorganisms for umami peptides formation. Functional annotation of five lactic acid bacteria indicated their important functions to be carbohydrate metabolism, amino acid metabolism and nucleotide metabolism, which proved their umami peptides production ability. Overall, our results enhanced the understanding of microbial communities and the formation mechanism of umami peptides in sufu, providing novel insights for quality control and flavor improvement of tofu products.

Toward a general and interpretable umami taste predictor using a multi-objective machine learning approach

The umami taste is one of the five basic taste modalities normally linked to the protein content in food. The implementation of fast and cost-effective tools for the prediction of the umami taste of a molecule remains extremely interesting to understand the molecular basis of this taste and to effectively rationalise the production and consumption of specific foods and ingredients. However, the only examples of umami predictors available in the literature rely on the amino acid sequence of the analysed peptides, limiting the applicability of the models. In the present study, we developed a novel ML-based algorithm, named VirtuousUmami, able to predict the umami taste of a query compound starting from its SMILES representation, thus opening up the possibility of potentially using such a model on any database through a standard and more general molecular description. Herein, we have tested our model on five databases related to foods or natural compounds. The proposed tool will pave the way toward the rationalisation of the molecular features underlying the umami taste and toward the design of specific peptide-inspired compounds with specific taste properties.

IUP-BERT: Identification of Umami Peptides Based on BERT Features

Umami is an important widely-used taste component of food seasoning. Umami peptides are specific structural peptides endowing foods with a favorable umami taste. Laboratory approaches used to identify umami peptides are time-consuming and labor-intensive, which are not feasible for rapid screening. Here, we developed a novel peptide sequence-based umami peptide predictor, namely iUP-BERT, which was based on the deep learning pretrained neural network feature extraction method. After optimization, a single deep representation learning feature encoding method (BERT: bidirectional encoder representations from transformer) in conjugation with the synthetic minority over-sampling technique (SMOTE) and support vector machine (SVM) methods was adopted for model creation to generate predicted probabilistic scores of potential umami peptides. Further extensive empirical experiments on cross-validation and an independent test showed that iUP-BERT outperformed the existing methods with improvements, highlighting its effectiveness and robustness. Finally, an open-access iUP-BERT web server was built. To our knowledge, this is the first efficient sequence-based umami predictor created based on a single deep-learning pretrained neural network feature extraction method. By predicting umami peptides, iUP-BERT can help in further research to improve the palatability of dietary supplements in the future.

DeepThal: A Deep Learning-Based Framework for the Large-Scale Prediction of the α+-Thalassemia Trait Using Red Blood Cell Parameters

Objectives: To develop a machine learning (ML)-based framework using red blood cell (RBC) parameters for the prediction of the α+-thalassemia trait (α+-thal trait) and to compare the diagnostic performance with a conventional method using a single RBC parameter or a combination of RBC parameters. Methods: A retrospective study was conducted on possible couples at risk for fetus with hemoglobin H (Hb H disease). Subjects with molecularly confirmed normal status (not thalassemia), α+-thal trait, and two-allele α-thalassemia mutation were included. Clinical parameters (age and gender) and RBC parameters (Hb, Hct, MCV, MCH, MCHC, RDW, and RBC count) obtained from their antenatal thalassemia screen were retrieved and analyzed using a machine learning (ML)-based framework and a conventional method. The performance of α+-thal trait prediction was evaluated. Results: In total, 594 cases (female/male: 330/264, mean age: 29.7 ± 6.6 years) were included in the analysis. There were 229 normal controls, 160 cases with the α+-thalassemia trait, and 205 cases in the two-allele α-thalassemia mutation category, respectively. The ML-derived model improved the diagnostic performance, giving a sensitivity of 80% and specificity of 81%. The experimental results indicated that DeepThal achieved a better performance compared with other ML-based methods in terms of the independent test dataset, with an accuracy of 80.77%, sensitivity of 70.59%, and the Matthews correlation coefficient (MCC) of 0.608. Of all the red blood cell parameters, MCH < 28.95 pg as a single parameter had the highest performance in predicting the α+-thal trait with the AUC of 0.857 and 95% CI of 0.816−0.899. The combination model derived from the binary logistic regression analysis exhibited improved performance with the AUC of 0.868 and 95% CI of 0.830−0.906, giving a sensitivity of 80.1% and specificity of 75.1%. Conclusions: The performance of DeepThal in terms of the independent test dataset is sufficient to demonstrate that DeepThal is capable of accurately predicting the α+-thal trait. It is anticipated that DeepThal will be a useful tool for the scientific community in the large-scale prediction of the α+-thal trait.

Taste GPCRs and their ligands

Taste GPCRs are expressed in taste buds on the tongue and play a key role in food choice and consumption. They are also expressed extra-orally, with various physiological roles that are currently under study. Unraveling the roles of these receptors relies on the knowledge of their ligands. Combining sensory, cell-based and computational approaches enabled the discovery of numerous agonists and several antagonists. Here we provide a short overview of taste receptor families, main recent methods for ligands discovery, and current sources of information about known ligands. The future directions that are likely to impact the taste GPCR field include focus on ligand interactions with naturally occurring polymorphisms, as well as harnessing the power of CryoEM and of multiple signaling readout techniques.

Computational prediction and interpretation of druggable proteins using a stacked ensemble-learning framework

Discovery of potential drugs requires rapid and precise identification of drug targets. Although traditional experimental methodologies can accurately identify drug targets, they are time-consuming and inappropriate for high-throughput screening. Computational approaches based on machine learning (ML) algorithms can expedite the prediction of druggable proteins; however, the performance of the existing computational methods remains unsatisfactory. This study proposes a computational tool, SPIDER, to enhance the accurate prediction of druggable proteins. SPIDER employs various feature descriptors pertaining to several aspects, including physicochemical properties, compositional information, and composition-transition-distribution information, coupled with well-known ML algorithms to facilitate the construction of the final meta-predictor. The experimental results showed that SPIDER enabled more precise and robust prediction of druggable proteins than the baseline models and current existing methods in terms of the independent test dataset. An online web server was established and made freely available online.

•

Computational models can expedite the identification of potential druggable proteins

•

SPIDER represents the first stacked model proposed for druggable protein prediction

•

SPIDER enables more precise prediction of druggable proteins than existing methods

•

The SPIDER web server is available at http://pmlabstack.pythonanywhere.com/SPIDER.

SAPPHIRE: A stacking-based ensemble learning framework for accurate prediction of thermophilic proteins

Thermophilic proteins (TPPs) are important in the field of protein biochemistry and development of new enzymes. Thus, computational methods must be urgently developed to accurately and rapidly identify TPPs. To date, several computational methods have been developed for TPP identification; however, few limitations in terms of performance and utility remain. In this study, we present a novel computational method, SAPPHIRE, to achieve more accurate identification of TPPs using only sequence information without any need for structural information. We combined twelve different feature encodings representing different perspectives and six popular machine learning algorithms to train 72 baseline models and extract the key information of TPPs. Subsequently, the informative predicted probabilities from the baseline models were mined and selected using a genetic algorithm in conjunction with a self-assessment-report approach. Finally, the final meta-predictor, SAPPHIRE, was built and optimized by applying an optimal feature set. The performance of SAPPHIRE in the 10-fold cross-validation test showed that a superior predictive performance compared with several baseline models could be achieved. Moreover, SAPPHIRE yielded an accuracy of 0.942 and Matthew's coefficient correlation of 0.884, which were 7.68 and 5.12% higher than those of the current existing methods, respectively, as indicated by the independent test. The proposed computational approach is anticipated to facilitate large-scale identification of TPPs and accelerate their applications in the food industry. The codes and datasets are available at https://github.com/plenoi/SAPPHIRE.

NEPTUNE: A novel computational approach for accurate and large-scale identification of tumor homing peptides

Tumor homing peptides (THPs) play a crucial role in recognizing and specifically binding to cancer cells. Although experimental approaches can facilitate the precise identification of THPs, they are usually time-consuming, labor-intensive, and not cost-effective. However, computational approaches can identify THPs by utilizing sequence information alone, thus highlighting their great potential for large-scale identification of THPs. Herein, we propose NEPTUNE, a novel computational approach for the accurate and large-scale identification of THPs from sequence information. Specifically, we constructed variant baseline models from multiple feature encoding schemes coupled with six popular machine learning algorithms. Subsequently, we comprehensively assessed and investigated the effects of these baseline models on THP prediction. Finally, the probabilistic information generated by the optimal baseline models is fed into a support vector machine-based classifier to construct the final meta-predictor (NEPTUNE). Cross-validation and independent tests demonstrated that NEPTUNE achieved superior performance for THP prediction compared with its constituent baseline models and the existing methods. Moreover, we employed the powerful SHapley additive exPlanations method to improve the interpretation of NEPTUNE and elucidate the most important features for identifying THPs. Finally, we implemented an online web server using NEPTUNE, which is available at http://pmlabstack.pythonanywhere.com/NEPTUNE. NEPTUNE could be beneficial for the large-scale identification of unknown THP candidates for follow-up experimental validation.

AMYPred-FRL is a novel approach for accurate prediction of amyloid proteins by using feature representation learning

Amyloid proteins have the ability to form insoluble fibril aggregates that have important pathogenic effects in many tissues. Such amyloidoses are prominently associated with common diseases such as type 2 diabetes, Alzheimer's disease, and Parkinson's disease. There are many types of amyloid proteins, and some proteins that form amyloid aggregates when in a misfolded state. It is difficult to identify such amyloid proteins and their pathogenic properties, but a new and effective approach is by developing effective bioinformatics tools. While several machine learning (ML)-based models for in silico identification of amyloid proteins have been proposed, their predictive performance is limited. In this study, we present AMYPred-FRL, a novel meta-predictor that uses a feature representation learning approach to achieve more accurate amyloid protein identification. AMYPred-FRL combined six well-known ML algorithms (extremely randomized tree, extreme gradient boosting, k-nearest neighbor, logistic regression, random forest, and support vector machine) with ten different sequence-based feature descriptors to generate 60 probabilistic features (PFs), as opposed to state-of-the-art methods developed by a single feature-based approach. A logistic regression recursive feature elimination (LR-RFE) method was used to find the optimal m number of 60 PFs in order to improve the predictive performance. Finally, using the meta-predictor approach, the 20 selected PFs were fed into a logistic regression method to create the final hybrid model (AMYPred-FRL). Both cross-validation and independent tests showed that AMYPred-FRL achieved superior predictive performance than its constituent baseline models. In an extensive independent test, AMYPred-FRL outperformed the existing methods by 5.5% and 16.1%, respectively, with accuracy and MCC of 0.873 and 0.710. To expedite high-throughput prediction, a user-friendly web server of AMYPred-FRL is freely available at http://pmlabstack.pythonanywhere.com/AMYPred-FRL. It is anticipated that AMYPred-FRL will be a useful tool in helping researchers to identify new amyloid proteins.