This study represents the first large-scale study on the chemical space of inhibitors of dipeptidyl peptidase-4 (DPP4), which is a potential therapeutic protein target for the treatment of diabetes mellitus. Herein, a large set of 2,937 compounds evaluated for their ability to inhibit DPP4 was compiled from the literature. Molecular descriptors were generated from the geometrically optimized low-energy conformers of these compounds at the semiempirical AM1 level. The origins of DPP4 inhibitory activity were elucidated from computed molecular descriptors that accounted for the unique physicochemical properties inherently present in the active and inactive sets of compounds as defined by their respective half maximal inhibitory concentration values of less than 1 μM and greater than 10 μM, respectively. Decision tree analysis revealed the importance of molecular weight, total energy of a molecule, topological polar surface area, lowest unoccupied molecular orbital, and number of hydrogen-bond donors, which correspond to molecular size, energy, surface polarity, electron acceptors, and hydrogen bond donors, respectively. The prediction model was subjected to rigorous independent testing via three external sets. Scaffold and chemical fragment analysis was also performed on these active and inactive sets of compounds to shed light on the distinguishing features of the functional moieties. Docking of representative active DPP4 inhibitors was also performed to unravel key interacting residues. The results of this study are anticipated to be useful in guiding the rational design of novel and robust DPP4 inhibitors for the treatment of diabetes.

P-glycoprotein (Pgp) is a drug transporter that plays important roles in multidrug resistance and drug pharmacokinetics. The inhibition of Pgp has become a notable strategy for combating multidrug-resistant cancers and improving therapeutic outcomes. However, the polyspecific nature of Pgp, together with inconsistent results in experimental assays, renders the determination of endpoints for Pgp-interacting compounds a great challenge. In this study, the classification of a large set of 2,477 Pgp-interacting compounds (i.e., 1341 inhibitors, 913 non-inhibitors, 197 substrates and 26 non-substrates) was performed using several machine learning methods (i.e., decision tree induction, artificial neural network modelling and support vector machine) as a function of their physicochemical properties. The models provided good predictive performance, producing MCC values in the range of 0.739-1 for internal cross-validation and 0.665-1 for external validation. The study provided simple and interpretable models for important properties that influence the activity of Pgp-interacting compounds, which are potentially beneficial for screening and rational design of Pgp inhibitors that are of clinical importance.

Understanding of quorum-sensing peptides (QSPs) in their functional mechanism plays an essential role in finding new opportunities to combat bacterial infections by designing drugs. With the avalanche of the newly available peptide sequences in the post-genomic age, it is highly desirable to develop a computational model for efficient, rapid and high-throughput QSP identification purely based on the peptide sequence information alone. Although, few methods have been developed for predicting QSPs, their prediction accuracy and interpretability still requires further improvements. Thus, in this work, we proposed an accurate sequence-based predictor (called iQSP) and a set of interpretable rules (called IR-QSP) for predicting and analyzing QSPs. In iQSP, we utilized a powerful support vector machine (SVM) cooperating with 18 informative features from physicochemical properties (PCPs). Rigorous independent validation test showed that iQSP achieved maximum accuracy and MCC of 93.00% and 0.86, respectively. Furthermore, a set of interpretable rules IR-QSP was extracted by using random forest model and the 18 informative PCPs. Finally, for the convenience of experimental scientists, the iQSP web server was established and made freely available online. It is anticipated that iQSP will become a useful tool or at least as a complementary existing method for predicting and analyzing QSPs.

Cancer imposes a global health burden as it represents one of the leading causes of morbidity and mortality while also giving rise to significant economic burden owing to the associated expenditures for its monitoring and treatment. In spite of advancements in cancer therapy, the low success rate and recurrence of tumor has necessitated the ongoing search for new therapeutic agents. Aside from drugs based on small molecules and protein-based biopharmaceuticals, there has been an intense effort geared towards the development of peptide-based therapeutics owing to its favorable and intrinsic properties of being relatively small, highly selective, potent, safe and low in production costs. In spite of these advantages, there are several inherent weaknesses that are in need of attention in the design and development of therapeutic peptides. An abundance of data on bioactive and therapeutic peptides have been accumulated over the years and the burgeoning area of artificial intelligence has set the stage for the lucrative utilization of machine learning to make sense of these large and high-dimensional data. This review summarizes the current state-of-the-art on the application of machine learning for studying the bioactivity of anticancer peptides along with future outlook of the field. Data and R codes used in the analysis herein are available on GitHub at https://github.com/Shoombuatong2527/anticancer-peptides-review.

Cancer remains one of the major causes of death worldwide. Angiogenesis is crucial for the pathogenesis of various human diseases, especially solid tumors. The discovery of anti-angiogenic peptides is a promising therapeutic route for cancer treatment. Thus, reliably identifying anti-angiogenic peptides is extremely important for understanding their biophysical and biochemical properties that serve as the basis for the discovery of new anti-cancer drugs. This study aims to develop an efficient and interpretable computational model called TargetAntiAngio for predicting and characterizing anti-angiogenic peptides. TargetAntiAngio was developed using the random forest classifier in conjunction with various classes of peptide features. It was observed via an independent validation test that TargetAntiAngio can identify anti-angiogenic peptides with an average accuracy of 77.50% on an objective benchmark dataset. Comparisons demonstrated that TargetAntiAngio is superior to other existing methods. In addition, results revealed the following important characteristics of anti-angiogenic peptides: (i) disulfide bond forming Cys residues play an important role for inhibiting blood vessel proliferation; (ii) Cys located at the C-terminal domain can decrease endothelial formatting activity and suppress tumor growth; and (iii) Cyclic disulfide-rich peptides contribute to the inhibition of angiogenesis and cell migration, selectivity and stability. Finally, for the convenience of experimental scientists, the TargetAntiAngio web server was established and made freely available online.

N7-methylguanosine (m7G) is an essential, ubiquitous, and positively charged modification at the 5' cap of eukaryotic mRNA, modulating its export, translation, and splicing processes. Although several machine learning (ML)-based computational predictors for m7G have been developed, all utilized specific computational framework. This study is the first instance we explored four different computational frameworks and identified the best approach. Based on that we developed a novel predictor, THRONE (A three-layer ensemble predictor for identifying human RNA N7-methylguanosine sites) to accurately identify m7G sites from the human genome. THRONE employs a wide range of sequence-based features inputted to several ML classifiers and combines these models through ensemble learning. The three-step ensemble learning is as follows: 54 baseline models were constructed in the first layer and the predicted probability of m7G was considered as a new feature vector for the sequential step. Subsequently, six meta-models were created using the new feature vector and their predicted probability was yet again considered as novel features. Finally, random forest was deemed as the best super classifier learner for the final prediction using a systematic approach incorporated with novel features. Interestingly, THRONE outperformed other existing methods in the prediction of m7G sites on both cross-validation analysis and independent evaluation. The proposed method is publicly accessible at: http://thegleelab.org/THRONE/ and expects to help the scientific community identify the putative m7G sites and formulate a novel testable biological hypothesis.

Anticancer peptides (ACPs) have emerged as a new class of therapeutic agent for cancer treatment due to their lower toxicity as well as greater efficacy, selectivity and specificity when compared to conventional small molecule drugs. However, the experimental identification of ACPs still remains a time-consuming and expensive endeavor. Therefore, it is desirable to develop and improve upon existing computational models for predicting and characterizing ACPs. In this study, we present a bioinformatics tool called the ACPred, which is an interpretable tool for the prediction and characterization of the anticancer activities of peptides. ACPred was developed by utilizing powerful machine learning models (support vector machine and random forest) and various classes of peptide features. It was observed by a jackknife cross-validation test that ACPred can achieve an overall accuracy of 95.61% in identifying ACPs. In addition, analysis revealed the following distinguishing characteristics that ACPs possess: (i) hydrophobic residue enhances the cationic properties of α-helical ACPs resulting in better cell penetration; (ii) the amphipathic nature of the α-helical structure plays a crucial role in its mechanism of cytotoxicity; and (iii) the formation of disulfide bridges on β-sheets is vital for structural maintenance which correlates with its ability to kill cancer cells. Finally, for the convenience of experimental scientists, the ACPred web server was established and made freely available online.

Currently, monomeric fluorescent proteins (FP) are ideal markers for protein tagging. The prediction of oligomeric states is helpful for enhancing live biomedical imaging. Computational prediction of FP oligomeric states can accelerate the effort of protein engineering efforts of creating monomeric FPs. To the best of our knowledge, this study represents the first computational model for predicting and analyzing FP oligomerization directly from the amino acid sequence.

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.

The development of efficient and effective bioinformatics tools and pipelines for identifying peptides with dipeptidyl peptidase IV (DPP-IV) inhibitory activities from large-scale protein datasets is of great importance for the discovery and development of potential and promising antidiabetic drugs. In this study, we present a novel stacking-based ensemble learning predictor (termed StackDPPIV) designed for identification of DPP-IV inhibitory peptides. Unlike the existing method, which is based on single-feature-based methods, we combined five popular machine learning algorithms in conjunction with ten different feature encodings from multiple perspectives to generate a pool of various baseline models. Subsequently, the probabilistic features derived from these baseline models were systematically integrated and deemed as new feature representations. Finally, in order to improve the predictive performance, the genetic algorithm based on the self-assessment-report was utilized to determine a set of informative probabilistic features and then used the optimal one for developing the final meta-predictor (StackDPPIV). Experiment results demonstrated that StackDPPIV could outperform its constituent baseline models on both the training and independent datasets. Furthermore, StackDPPIV achieved an accuracy of 0.891, MCC of 0.784 and AUC of 0.961, which were 9.4%, 19.0% and 11.4%, respectively, higher than that of the existing method on the independent test. Feature analysis demonstrated that our feature representations had more discriminative ability as compared to conventional feature descriptors, which highlights the combination of different features was essential for the performance improvement. In order to implement the proposed predictor, we had built a user-friendly online web server at http://pmlabstack.pythonanywhere.com/StackDPPIV.

Pupylation is a type of reversible post-translational modification of proteins, which plays a key role in the cellular function of microbial organisms. Several proteomics methods have been developed for the prediction and analysis of pupylated proteins and pupylation sites. However, the traditional experimental methods are laborious and time-consuming. Hence, computational algorithms are highly needed that can predict potential pupylation sites using sequence features. In this research, a new prediction model, PUP-Fuse, has been developed for pupylation site prediction by integrating multiple sequence representations. Meanwhile, we explored the five types of feature encoding approaches and three machine learning (ML) algorithms. In the final model, we integrated the successive ML scores using a linear regression model. The PUP-Fuse achieved a Mathew correlation value of 0.768 by a 10-fold cross-validation test. It also outperformed existing predictors in an independent test. The web server of the PUP-Fuse with curated datasets is freely available.

Progesterone receptors (PRs) are implicated in various cancers since their presence/absence can determine clinical outcomes. The overstimulation of progesterone can facilitate oncogenesis and thus, its modulation through PR inhibition is urgently needed. To address this issue, a novel stacked ensemble learning approach (termed StackPR) is presented for fast, accurate, and large-scale identification of PR antagonists using only SMILES notation without the need for 3D structural information. We employed six popular machine learning (ML) algorithms (i.e., logistic regression, partial least squares, k-nearest neighbor, support vector machine, extremely randomized trees, and random forest) coupled with twelve conventional molecular descriptors to create 72 baseline models. Then, a genetic algorithm in conjunction with the self-assessment-report approach was utilized to determine m out of the 72 baseline models as means of developing the final meta-predictor using the stacking strategy and tenfold cross-validation test. Experimental results on the independent test dataset show that StackPR achieved impressive predictive performance with an accuracy of 0.966 and Matthew's coefficient correlation of 0.925. In addition, analysis based on the SHapley Additive exPlanation algorithm and molecular docking indicates that aliphatic hydrocarbons and nitrogen-containing substructures were the most important features for having PR antagonist activity. Finally, we implemented an online webserver using StackPR, which is freely accessible at http://pmlabstack.pythonanywhere.com/StackPR . StackPR is anticipated to be a powerful computational tool for the large-scale identification of unknown PR antagonist candidates for follow-up experimental validation.

Alzheimer's disease (AD) is a chronic neurodegenerative disease which leads to the gradual loss of neuronal cells. Several hypotheses for AD exists (e.g., cholinergic, amyloid, tau hypotheses, etc.). As per the cholinergic hypothesis, the deficiency of choline is responsible for AD; therefore, the inhibition of AChE is a lucrative therapeutic strategy for the treatment of AD. Acetylcholinesterase (AChE) is an enzyme that catalyzes the breakdown of the neurotransmitter acetylcholine that is essential for cognition and memory. A large non-redundant data set of 2,570 compounds with reported IC50 values against AChE was obtained from ChEMBL and employed in quantitative structure-activity relationship (QSAR) study so as to gain insights on their origin of bioactivity. AChE inhibitors were described by a set of 12 fingerprint descriptors and predictive models were constructed from 100 different data splits using random forest. Generated models afforded R (2), [Formula: see text] and [Formula: see text] values in ranges of 0.66-0.93, 0.55-0.79 and 0.56-0.81 for the training set, 10-fold cross-validated set and external set, respectively. The best model built using the substructure count was selected according to the OECD guidelines and it afforded R (2), [Formula: see text] and [Formula: see text] values of 0.92 ± 0.01, 0.78 ± 0.06 and 0.78 ± 0.05, respectively. Furthermore, Y-scrambling was applied to evaluate the possibility of chance correlation of the predictive model. Subsequently, a thorough analysis of the substructure fingerprint count was conducted to provide informative insights on the inhibitory activity of AChE inhibitors. Moreover, Kennard-Stone sampling of the actives were applied to select 30 diverse compounds for further molecular docking studies in order to gain structural insights on the origin of AChE inhibition. Site-moiety mapping of compounds from the diversity set revealed three binding anchors encompassing both hydrogen bonding and van der Waals interaction. Molecular docking revealed that compounds 13, 5 and 28 exhibited the lowest binding energies of -12.2, -12.0 and -12.0 kcal/mol, respectively, against human AChE, which is modulated by hydrogen bonding, π-π stacking and hydrophobic interaction inside the binding pocket. These information may be used as guidelines for the design of novel and robust AChE inhibitors.

The fight against the emergence of mutant influenza strains has led to the screening of an increasing number of compounds for inhibitory activity against influenza neuraminidase. This study explores the chemical space of neuraminidase inhibitors (NAIs), which provides an opportunity to obtain further molecular insights regarding the underlying basis of their bioactivity. In particular, a large set of 347 and 175 NAIs against influenza A and B, respectively, was compiled from the literature. Molecular and quantum chemical descriptors were obtained from low-energy conformational structures geometrically optimized at the PM6 level. The bioactivities of NAIs were classified as active or inactive according to their half maximum inhibitory concentration (IC50) value in which IC50 < 1µM and ≥ 10µM were defined as active and inactive compounds, respectively. Interpretable decision rules were derived from a quantitative structure-activity relationship (QSAR) model established using a set of substructure descriptors via decision tree analysis. Univariate analysis, feature importance analysis from decision tree modeling and molecular scaffold analysis were performed on both data sets for discriminating important structural features amongst active and inactive NAIs. Good predictive performance was achieved as deduced from accuracy and Matthews correlation coefficient values in excess of 81% and 0.58, respectively, for both influenza A and B NAIs. Furthermore, molecular docking was employed to investigate the binding modes and their moiety preferences of active NAIs against both influenza A and B neuraminidases. Moreover, novel NAIs with robust binding fitness towards influenza A and B neuraminidase were generated via combinatorial library enumeration and their binding fitness was on par or better than FDA-approved drugs. The results from this study are anticipated to be beneficial for guiding the rational drug design of novel NAIs for treating influenza infections.

Background: Dyslipidemia is one of the major forms of lipid disorder, characterized by increased triglycerides (TGs), increased low-density lipoprotein-cholesterol (LDL-C), and decreased high-density lipoprotein-cholesterol (HDL-C) levels in blood. Recently, MicroRNAs (miRNAs) have been reported to involve in various biological processes; their potential usage being a biomarkers and in diagnosis of various diseases. Computational approaches including text mining have been used recently to analyze abstracts from the public databases to observe the relationships/associations between the biological molecules, miRNAs, and disease phenotypes. Materials and Methods: In the present study, significance of text mined extracted pair associations (miRNA-lipid disease) were estimated by one-sided Fisher's exact test. The top 20 significant miRNA-disease associations were visualized on Cytoscape. The CyTargetLinker plug-in tool on Cytoscape was used to extend the network and predicts new miRNA target genes. The Biological Networks Gene Ontology (BiNGO) plug-in tool on Cytoscape was used to retrieve gene ontology (GO) annotations for the targeted genes. Results: We retrieved 227 miRNA-lipid disease associations including 148 miRNAs. The top 20 significant miRNAs analysis on CyTargetLinker provides defined, predicted and validated gene targets, further targeted genes analyzed by BiNGO showed targeted genes were significantly associated with lipid, cholesterol, apolipoprotein, and fatty acids GO terms. Conclusion: We are the first to provide a reliable miRNA-lipid disease association network based on text mining. This could help future experimental studies that aim to validate predicted gene targets.

In spite of the large-scale production and widespread distribution of vaccines and antiviral drugs, viruses remain a prominent human disease. Recently, the discovery of antiviral peptides (AVPs) has become an influential antiviral agent due to their extraordinary advantages. With the avalanche of newly-found peptide sequences in the post-genomic era, there is a great demand to develop a sequence-based predictor for timely identifying AVPs as this information is very useful for both basic research and drug development. In this study, we propose a novel sequence-based meta-predictor with an effective feature representation, called Meta-iAVP, for the accurate prediction of AVPs from given peptide sequences. Herein, the effective feature representation was extracted from a set of prediction scores derived from various machine learning algorithms and types of features. To the best of our knowledge, the model proposed herein represents the first meta-based approach for the prediction of AVPs. An overall accuracy and Matthews correlation coefficient of 95.20% and 0.90, respectively, was achieved from the independent test set on an objective benchmark dataset. Comparative analysis suggested that Meta-iAVP was superior to that of existing methods and therefore represents a useful tool for AVP prediction. Finally, in an effort to facilitate high-throughput prediction of AVPs, the model was deployed as the Meta-iAVP web server and is made freely available online at http://codes.bio/meta-iavp/ where users can submit query peptide sequences for determining the likelihood of whether or not these peptides are AVPs.

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.

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.

Owing to their ability to maintain a thermodynamically stable fold at extremely high temperatures, thermophilic proteins (TTPs) play a critical role in basic research and a variety of applications in the food industry. As a result, the development of computation models for rapidly and accurately identifying novel TTPs from a large number of uncharacterized protein sequences is desirable. In spite of existing computational models that have already been developed for characterizing thermophilic proteins, their performance and interpretability remain unsatisfactory. We present a novel sequence-based thermophilic protein predictor, termed SCMTPP, for improving model predictability and interpretability. First, an up-to-date and high-quality dataset consisting of 1853 TPPs and 3233 non-TPPs was compiled from published literature. Second, the SCMTPP predictor was created by combining the scoring card method (SCM) with estimated propensity scores of g-gap dipeptides. Benchmarking experiments revealed that SCMTPP had a cross-validation accuracy of 0.883, which was comparable to that of a support vector machine-based predictor (0.906-0.910) and 2-17% higher than that of commonly used machine learning models. Furthermore, SCMTPP outperformed the state-of-the-art approach (ThermoPred) on the independent test dataset, with accuracy and MCC of 0.865 and 0.731, respectively. Finally, the SCMTPP-derived propensity scores were used to elucidate the critical physicochemical properties for protein thermostability enhancement. In terms of interpretability and generalizability, comparative results showed that SCMTPP was effective for identifying and characterizing TPPs. We had implemented the proposed predictor as a user-friendly online web server at http://pmlabstack.pythonanywhere.com/SCMTPP in order to allow easy access to the model. SCMTPP is expected to be a powerful tool for facilitating community-wide efforts to identify TPPs on a large scale and guiding experimental characterization of TPPs.

N4-methylcytosine (4mC) is one of the most important DNA modifications and involved in regulating cell differentiations and gene expressions. The accurate identification of 4mC sites is necessary to understand various biological functions. In this work, we developed a new computational predictor called i4mC-Mouse to identify 4mC sites in the mouse genome. Herein, six encoding schemes of k-space nucleotide composition (KSNC), k-mer nucleotide composition (Kmer), mono nucleotide binary encoding (MBE), dinucleotide binary encoding, electron-ion interaction pseudo potentials (EIIP) and dinucleotide physicochemical composition were explored that cover different characteristics of DNA sequence information. Subsequently, we built six RF-based encoding models and then linearly combined their probability scores to construct the final predictor. Among the six RF-based models, the Kmer, KSNC, MBE, and EIIP encodings are sufficient, which contributed to 10%, 45%, 25%, and 20% of the prediction performance, respectively. On the independent test the i4mC-Mouse predicted the 4mC sites with accuracy and MCC of 0.816 and 0.633, respectively, which were approximately 2.5% and 5% higher than those of the existing method (4mCpred-EL). For experimental biologists, a freely available web application was implemented at http://kurata14.bio.kyutech.ac.jp/i4mC-Mouse/.