Significantly improved HIV inhibitor efficacy prediction employing proteochemometric models generated from antivirogram data

A study of the interaction space of two lactate dehydrogenase isoforms (LDHA and LDHB) and some of their inhibitors using proteochemometrics modeling

Lactate dehydrogenase (LDH) is a tetramer enzyme that converts pyruvate to lactate reversibly. This enzyme becomes important because it is associated with diseases such as cancers, heart disease, liver problems, and most importantly, corona disease. As a system-based method, proteochemometrics does not require knowledge of the protein's three-dimensional structure, but rather depends on the amino acid sequence and protein descriptors. Here, we applied this methodology to model a set of LDHA and LDHB isoenzyme inhibitors. To implement the proteochemetrics method, the camb package in the R Studio Server programming environment was used. The activity of 312 compounds of LDHA and LDHB isoenzyme inhibitors from the valid Binding DB database was retrieved. The proteochemometrics method was applied to three machine learning algorithms gradient amplification model, random forest, and support vector machine as regression methods to find the best model. Through the combination of different models into an ensemble (greedy and stacking optimization), we explored the possibility of improving the performance of models. For the RF best ensemble model of inhibitors of LDHA and LDHB isoenzymes, and were 0.66 and 0.62, respectively. LDH inhibitory activation is influenced by Morgan fingerprints and topological structure descriptors.

Proteochemometric Modeling Identifies Chemically Diverse Norepinephrine Transporter Inhibitors

Solute carriers (SLCs) are relatively underexplored compared to other prominent protein families such as kinases and G protein-coupled receptors. However, proteins from the SLC family play an essential role in various diseases. One such SLC is the high-affinity norepinephrine transporter (NET/SLC6A2). In contrast to most other SLCs, the NET has been relatively well studied. However, the chemical space of known ligands has a low chemical diversity, making it challenging to identify chemically novel ligands. Here, a computational screening pipeline was developed to find new NET inhibitors. The approach increases the chemical space to model for NETs using the chemical space of related proteins that were selected utilizing similarity networks. Prior proteochemometric models added data from related proteins, but here we use a data-driven approach to select the optimal proteins to add to the modeled data set. After optimizing the data set, the proteochemometric model was optimized using stepwise feature selection. The final model was created using a two-step approach combining several proteochemometric machine learning models through stacking. This model was applied to the extensive virtual compound database of Enamine, from which the top predicted 22,000 of the 600 million virtual compounds were clustered to end up with 46 chemically diverse candidates. A subselection of 32 candidates was synthesized and subsequently tested using an impedance-based assay. There were five hit compounds identified (hit rate 16%) with sub-micromolar inhibitory potencies toward NET, which are promising for follow-up experimental research. This study demonstrates a data-driven approach to diversify known chemical space to identify novel ligands and is to our knowledge the first to select this set based on the sequence similarity of related targets.

Artificial intelligence and machine learning assisted drug delivery for effective treatment of infectious diseases

In the era of antimicrobial resistance, the prevalence of multidrug-resistant microorganisms that resist conventional antibiotic treatment has steadily increased. Thus, it is now unquestionable that infectious diseases are significant global burdens that urgently require innovative treatment strategies. Emerging studies have demonstrated that artificial intelligence (AI) can transform drug delivery to promote effective treatment of infectious diseases. In this review, we propose to evaluate the significance, essential principles, and popular tools of AI in drug delivery for infectious disease treatment. Specifically, we will focus on the achievements and key findings of current research, as well as the applications of AI on drug delivery throughout the whole antimicrobial treatment process, with an emphasis on drug development, treatment regimen optimization, drug delivery system and administration route design, and drug delivery outcome prediction. To that end, the challenges of AI in drug delivery for infectious disease treatments and their current solutions and future perspective will be presented and discussed.

Revealing the Mutation Patterns of Drug-Resistant Reverse Transcriptase Variants of Human Immunodeficiency Virus through Proteochemometric Modeling

Drug-resistant cases of human immunodeficiency virus (HIV) nucleoside reverse transcriptase inhibitors (NRTI) are constantly accumulating due to the frequent mutations of the reverse transcriptase (RT). Predicting the potential drug resistance of HIV-1 NRTIs could provide instructions for the proper clinical use of available drugs. In this study, a novel proteochemometric (PCM) model was constructed to predict the drug resistance between six NRTIs against different variants of RT. Forty-seven dominant mutation sites were screened using the whole protein of HIV-1 RT. Thereafter, the physicochemical properties of the dominant mutation sites can be derived to generate the protein descriptors of RT. Furthermore, by combining the molecular descriptors of NRTIs, PCM modeling can be constructed to predict the inhibition ability between RT variants and NRTIs. The results indicated that our PCM model could achieve a mean AUC value of 0.946 and a mean accuracy of 0.873 on the external validation set. Finally, based on PCM modeling, the importance of features was calculated to reveal the dominant amino acid distribution and mutation patterns on RT, to reflect the characteristics of drug-resistant sequences.

Machine Learning in Discovery of New Antivirals and Optimization of Viral Infections Therapy

Nowadays, computational approaches play an important role in the design of new drug-like compounds and optimization of pharmacotherapeutic treatment of diseases. The emerging growth of viral infections, including those caused by the Human Immunodeficiency Virus (HIV), Ebola virus, recently detected coronavirus, and some others, leads to many newly infected people with a high risk of death or severe complications. A huge amount of chemical, biological, clinical data is at the disposal of the researchers. Therefore, there are many opportunities to find the relationships between the particular features of chemical data and the antiviral activity of biologically active compounds based on machine learning approaches. Biological and clinical data can also be used for building models to predict relationships between viral genotype and drug resistance, which might help determine the clinical outcome of treatment. In the current study, we consider machine-learning approaches in the antiviral research carried out during the past decade. We overview in detail the application of machine-learning methods for the design of new potential antiviral agents and vaccines, drug resistance prediction, and analysis of virus-host interactions. Our review also covers the perspectives of using the machine-learning approaches for antiviral research, including Dengue, Ebola viruses, Influenza A, Human Immunodeficiency Virus, coronaviruses, and some others.

A Computational Approach for the Prediction of Treatment History and the Effectiveness or Failure of Antiretroviral Therapy

Human Immunodeficiency Virus Type 1 (HIV-1) infection is associated with high mortality if no therapy is provided. Currently, the treatment of an HIV-1 positive patient requires that several drugs should be taken simultaneously. The resistance of the virus to an antiretroviral drug may lead to treatment failure. Our approach focuses on predicting the exposure of a particular viral variant to an antiretroviral drug or drug combination. It also aims at the prediction of drug treatment success or failure. We utilized nucleotide sequences of HIV-1 encoding protease and reverse transcriptase to perform such types of prediction. The PASS (Prediction of Activity Spectra for Substances) algorithm based on the naive Bayesian classifier was used to make a prediction. We calculated the probability of whether a sequence belonged (P₁) or did not belong (P₀) to the class associated with exposure of the viral sequence to the set of drugs that can be associated with resistance to the set of drugs. The accuracy calculated as the average Area Under the ROC (Receiver Operating Characteristic) Curve (AUC/ROC) for classifying exposure of the sequence to the HIV-1 protease inhibitors was 0.81 (±0.07), and for HIV-1 reverse transcriptase, it was 0.83 (±0.07). To predict cases of treatment effectiveness or failure, we used P₁ and P₀ values, obtained in PASS, along with the binary vector constructed based on short nucleotide descriptors and the applied random forest classifier. Average AUC/ROC prediction accuracy for the prediction of treatment effectiveness or failure for the combinations of HIV-1 protease inhibitors was 0.82 (±0.06) and of HIV-1 reverse transcriptase was 0.76 (±0.09).

Modeling cancer drug response through drug-specific informative genes

Recent advances in pharmacogenomics have generated a wealth of data of different types whose analysis have helped in the identification of signatures of different cellular sensitivity/resistance responses to hundreds of chemical compounds. Among the different data types, gene expression has proven to be the more successful for the inference of drug response in cancer cell lines. Although effective, the whole transcriptome can introduce noise in the predictive models, since specific mechanisms are required for different drugs and these realistically involve only part of the proteins encoded in the genome. We analyzed the pharmacogenomics data of 961 cell lines tested with 265 anti-cancer drugs and developed different machine learning approaches for dissecting the genome systematically and predict drug responses using both drug-unspecific and drug-specific genes. These methodologies reach better response predictions for the vast majority of the screened drugs using tens to few hundreds genes specific to each drug instead of the whole genome, thus allowing a better understanding and interpretation of drug-specific response mechanisms which are not necessarily restricted to the drug known targets.

Novel Protease Inhibitors Containing C-5-Modified bis-Tetrahydrofuranylurethane and Aminobenzothiazole as P2 and P2' Ligands That Exert Potent Antiviral Activity against Highly Multidrug-Resistant HIV-1 with a High Genetic Barrier against the Emergence of Drug Resistance

Combination antiretroviral therapy has achieved dramatic reductions in the mortality and morbidity in people with HIV-1 infection. Darunavir (DRV) represents a most efficacious and well-tolerated protease inhibitor (PI) with a high genetic barrier to the emergence of drug-resistant HIV-1. However, highly DRV-resistant variants have been reported in patients receiving long-term DRV-containing regimens. Here, we report three novel HIV-1 PIs (GRL-057-14, GRL-058-14, and GRL-059-14), all of which contain a P2-amino-substituted-bis-tetrahydrofuranylurethane (bis-THF) and a P2'-cyclopropyl-amino-benzothiazole (Cp-Abt). These PIs not only potently inhibit the replication of wild-type HIV-1 (50% effective concentration [EC₅₀], 0.22 nM to 10.4 nM) but also inhibit multi-PI-resistant HIV-1 variants, including highly DRV-resistant HIV_DRV ^R _P51 (EC₅₀, 1.6 nM to 30.7 nM). The emergence of HIV-1 variants resistant to the three compounds was much delayed in selection experiments compared to resistance to DRV, using a mixture of 11 highly multi-PI-resistant HIV-1 isolates as a starting HIV-1 population. GRL-057-14 showed the most potent anti-HIV-1 activity and greatest thermal stability with wild-type protease, and potently inhibited HIV-1 protease's proteolytic activity (K_i value, 0.10 nM) among the three PIs. Structural models indicate that the C-5-isopropylamino-bis-THF moiety of GRL-057-14 forms additional polar interactions with the active site of HIV-1 protease. Moreover, GRL-057-14's P1-bis-fluoro-methylbenzene forms strong hydrogen bonding and effective van der Waals interactions. The present data suggest that the combination of C-5-aminoalkyl-bis-THF, P1-bis-fluoro-methylbenzene, and P2'-Cp-Abt confers highly potent activity against wild-type and multi-PI-resistant HIV strains and warrant further development of the three PIs, in particular, that of GRL-057-14, as potential therapeutic for HIV-1 infection and AIDS.

Mapping the drivers of within-host pathogen evolution using massive data sets

Differences among hosts, resulting from genetic variation in the immune system or heterogeneity in drug treatment, can impact within-host pathogen evolution. Genetic association studies can potentially identify such interactions. However, extensive and correlated genetic population structure in hosts and pathogens presents a substantial risk of confounding analyses. Moreover, the multiple testing burden of interaction scanning can potentially limit power. We present a Bayesian approach for detecting host influences on pathogen evolution that exploits vast existing data sets of pathogen diversity to improve power and control for stratification. The approach models key processes, including recombination and selection, and identifies regions of the pathogen genome affected by host factors. Our simulations and empirical analysis of drug-induced selection on the HIV-1 genome show that the method recovers known associations and has superior precision-recall characteristics compared to other approaches. We build a high-resolution map of HLA-induced selection in the HIV-1 genome, identifying novel epitope-allele combinations.

[Computational prediction of human immunodeficiency resistance to reverse transcriptase inhibitors]

Human immunodeficiency virus (HIV) causes acquired immunodeficiency syndrome (AIDS) and leads to over one million of deaths annually. Highly active antiretroviral treatment (HAART) is a gold standard in the HIV/AIDS therapy. Nucleoside and non-nucleoside inhibitors of HIV reverse transcriptase (RT) are important component of HAART, but their effect depends on the HIV susceptibility/resistance. HIV resistance mainly occurs due to mutations leading to conformational changes in the three-dimensional structure of HIV RT. The aim of our work was to develop and test a computational method for prediction of HIV resistance associated with the mutations in HIV RT. Earlier we have developed a method for prediction of HIV type 1 (HIV-1) resistance; it is based on the usage of position-specific descriptors. These descriptors are generated using the particular amino acid residue and its position; the position of certain residue is determined in a multiple alignment. The training set consisted of more than 1900 sequences of HIV RT from the Stanford HIV Drug Resistance database; for these HIV RT variants experimental data on their resistance to ten inhibitors are presented. Balanced accuracy of prediction varies from 80% to 99% depending on the method of classification (support vector machine, Naive Bayes, random forest, convolutional neural networks) and the drug, resistance to which is obtained. Maximal balanced accuracy was obtained for prediction of resistance to zidovudine, stavudine, didanosine and efavirenz by the random forest classifier. Average accuracy of prediction is 89%.

Current Advances in Studying Clinically Relevant Transporters of the Solute Carrier (SLC) Family by Connecting Computational Modeling and Data Science

Organic anion and cation transporting proteins (OATs, OATPs, and OCTs), as well as the Multidrug and Toxin Extrusion (MATE) transporters of the Solute Carrier (SLC) family are playing a pivotal role in the discovery and development of new drugs due to their involvement in drug disposition, drug-drug interactions, adverse drug effects and related toxicity. Computational methods to understand and predict clinically relevant transporter interactions can provide useful guidance at early stages in drug discovery and design, especially if they include contemporary data science approaches. In this review, we summarize the current state-of-the-art of computational approaches for exploring ligand interactions and selectivity for these drug (uptake) transporters. The computational methods discussed here by highlighting interesting examples from the current literature are ranging from semiautomatic data mining and integration, to ligand-based methods (such as quantitative structure-activity relationships, and combinatorial pharmacophore modeling), and finally structure-based methods (such as comparative modeling, molecular docking, and molecular dynamics simulations). We are focusing on promising computational techniques such as fold-recognition methods, proteochemometric modeling or techniques for enhanced sampling of protein conformations used in the context of these ADMET-relevant SLC transporters with a special focus on methods useful for studying ligand selectivity.

Nucleic acid testing and molecular characterization of HIV infections

Significant advances have been made in the molecular assays used for the detection of human immunodeficiency virus (HIV), which are crucial in preventing HIV transmission and monitoring disease progression. Molecular assays for HIV diagnosis have now reached a high degree of specificity, sensitivity and reproducibility, and have less operator involvement to minimize risk of contamination. Furthermore, analyses have been developed for the characterization of host gene polymorphisms and host responses to better identify and monitor HIV-1 infections in the clinic. Currently, molecular technologies including HIV quantitative and qualitative assays are mainly based on the polymerase chain reaction (PCR), transcription-mediated amplification (TMA), nucleic acid sequence-based amplification (NASBA), and branched chain (b) DNA methods and widely used for HIV detection and characterization, such as blood screening, point-of-care testing (POCT), pediatric diagnosis, acute HIV infection (AHI), HIV drug resistance testing, antiretroviral (AR) susceptibility testing, host genome polymorphism testing, and host response analysis. This review summarizes the development and the potential utility of molecular assays used to detect and characterize HIV infections.

A Computational Approach for the Prediction of HIV Resistance Based on Amino Acid and Nucleotide Descriptors

The high variability of the human immunodeficiency virus (HIV) is an important cause of HIV resistance to reverse transcriptase and protease inhibitors. There are many variants of HIV type 1 (HIV-1) that can be used to model sequence-resistance relationships. Machine learning methods are widely and successfully used in new drug discovery. An emerging body of data regarding the interactions of small drug-like molecules with their protein targets provides the possibility of building models on "structure-property" relationships and analyzing the performance of various machine-learning techniques. In our research, we analyze several different types of descriptors in order to predict the resistance of HIV reverse transcriptase and protease to the marketed antiretroviral drugs using the Random Forest approach. First, we represented amino acid sequences as a set of short peptide fragments, which included several amino acid residues. Second, we represented nucleotide sequences as a set of fragments, which included several nucleotides. We compared these two approaches using open data from the Stanford HIV Drug Resistance Database. We have determined the factors that modulate the performance of prediction: in particular, we observed that the prediction performance was more sensitive to certain drugs than a type of the descriptor used.

HIV Resistance Prediction to Reverse Transcriptase Inhibitors: Focus on Open Data

Research and development of new antiretroviral agents are in great demand due to issues with safety and efficacy of the antiretroviral drugs. HIV reverse transcriptase (RT) is an important target for HIV treatment. RT inhibitors targeting early stages of the virus-host interaction are of great interest for researchers. There are a lot of clinical and biochemical data on relationships between the occurring of the single point mutations and their combinations in the pol gene of HIV and resistance of the particular variants of HIV to nucleoside and non-nucleoside reverse transcriptase inhibitors. The experimental data stored in the databases of HIV sequences can be used for development of methods that are able to predict HIV resistance based on amino acid or nucleotide sequences. The data on HIV sequences resistance can be further used for (1) development of new antiretroviral agents with high potential for HIV inhibition and elimination and (2) optimization of antiretroviral therapy. In our communication, we focus on the data on the RT sequences and HIV resistance, which are available on the Internet. The experimental methods, which are applied to produce the data on HIV-1 resistance, the known data on their concordance, are also discussed.

Identification of Allosteric Modulators of Metabotropic Glutamate 7 Receptor Using Proteochemometric Modeling

Proteochemometric modeling (PCM) is a computational approach that can be considered an extension of quantitative structure-activity relationship (QSAR) modeling, where a single model incorporates information for a family of targets and all the associated ligands instead of modeling activity versus one target. This is especially useful for situations where bioactivity data exists for similar proteins but is scarce for the protein of interest. Here we demonstrate the application of PCM to identify allosteric modulators of metabotropic glutamate (mGlu) receptors. Given our long-running interest in modulating mGlu receptor function we compiled a matrix of compound-target bioactivity data. Some members of the mGlu family are well explored both internally and in the public domain, while there are much fewer examples of ligands for other targets such as the mGlu7 receptor. Using a PCM approach mGlu7 receptor hits were found. In comparison to conventional single target modeling the identified hits were more diverse, had a better confirmation rate, and provide starting points for further exploration. We conclude that the robust structure-activity relationship from well explored target family members translated to better quality hits for PCM compared to virtual screening (VS) based on a single target.

Conformations of the HIV-1 protease: A crystal structure data set analysis

The HIV protease is an important drug target for HIV/AIDS therapy, and its structure and function have been extensively investigated. This enzyme performs an essential role in viral maturation by processing specific cleavage sites in the Gag and Gag-Pol precursor polyproteins so as to release their mature forms. This 99 amino acid aspartic protease works as a homodimer, with the active site localized in a central cavity capped by two flexible flap regions. The dimer presents closed or open conformations, which are involved in the substrate binding and release. Here the results of the analysis of a HIV-1 protease data set containing 552 dimer structures are reported. Different dimensionality reduction methods have been used in order to get information from this multidimensional database. Most of the structures in the data set belong to two conformational clusters. An interesting observation that comes from the analysis of these data is that some protease sequences are localized preferentially in specific areas of the conformational landscape of this protein.

HIV drug resistance in HIV positive individuals under antiretroviral treatment in Shandong Province, China

The efficacy of antiretroviral drugs is limited by the development of drug resistance. Therefore, it is important to examine HIV drug resistance following the nationwide implementation of drug resistance testing in China since 2009. We conducted drug resistance testing in patients who were already on or new to HIV antiretroviral therapy (ART) in Shandong Province, China, from 2011 to 2013, and grouped them based on the presence or absence of drug resistance to determine the effects of age, gender, ethnicity, marital status, educational level, route of transmission and treatment status on drug resistance. We then examined levels of drug resistance the following year. The drug resistance rates of HIV patients on ART in Shandong from 2011 to 2013 were 3.45% (21/608), 3.38% (31/916), and 4.29% (54/1259), per year, respectively. M184V was the most frequently found point mutation, conferring resistance to the nucleoside reverse transcriptase inhibitor, while Y181C, G190A, K103N and V179D/E/F were the most frequent point mutations conferring resistance to the non-nucleoside reverse transcriptase inhibitor. In addition, the protease inhibitor drug resistance mutations I54V and V82A were identified for the first time in Shandong Province. Primary resistance accounts for 20% of the impact factors for drug resistance. Furthermore, it was found that educational level and treatment regimen were high-risk factors for drug resistance in 2011 (P<0.05), while treatment regimen was a high risk factor for drug resistance in 2012 and 2013 (P<0.05). Among the 106 drug-resistant patients, 77 received immediate adjustment of treatment regimen following testing, and 69 (89.6%) showed a reduction in drug resistance the following year. HIV drug resistance has a low prevalence in Shandong Province. However, patients on second line ART regimens and those with low educational level need continuous monitoring. Active drug resistance testing can effectively prevent the development of drug resistance.

A novel multi-target regression framework for time-series prediction of drug efficacy

Excavating from small samples is a challenging pharmacokinetic problem, where statistical methods can be applied. Pharmacokinetic data is special due to the small samples of high dimensionality, which makes it difficult to adopt conventional methods to predict the efficacy of traditional Chinese medicine (TCM) prescription. The main purpose of our study is to obtain some knowledge of the correlation in TCM prescription. Here, a novel method named Multi-target Regression Framework to deal with the problem of efficacy prediction is proposed. We employ the correlation between the values of different time sequences and add predictive targets of previous time as features to predict the value of current time. Several experiments are conducted to test the validity of our method and the results of leave-one-out cross-validation clearly manifest the competitiveness of our framework. Compared with linear regression, artificial neural networks, and partial least squares, support vector regression combined with our framework demonstrates the best performance, and appears to be more suitable for this task.

PASS-based approach to predict HIV-1 reverse transcriptase resistance

HIV reverse transcriptase (RT) inhibitors targeting the early stages of virus-host interactions are of great interest to scientists. Acquired HIV RT resistance happens due to mutations in a particular region of the pol gene encoding the HIV RT amino acid sequence. We propose an application of the previously developed PASS algorithm for prediction of amino acid substitutions potentially involved in the resistance of HIV-1 based on open data. In our work, we used more than 3200 HIV-1 RT variants from the publicly available Stanford HIV RT and protease sequence database already tested for 10 anti-HIV drugs including both nucleoside and non-nucleoside RT inhibitors. We used a particular amino acid residue and its position to describe primary structure-resistance relationships. The average balanced accuracy of the prediction obtained in 20-fold cross-validation for the Phenosense dataset was about 88% and for the Antivirogram dataset was about 79%. Thus, the PASS-based algorithm may be used for prediction of the amino acid substitutions associated with the resistance of HIV-1 based on open data. The computational approach for the prediction of HIV-1 associated resistance can be useful for the selection of RT inhibitors for the treatment of HIV infected patients in the clinical practice. Prediction of the HIV-1 RT associated resistance can be useful for the development of new anti-HIV drugs active against the resistant variants of RT. Therefore, we propose that this study can be potentially useful for anti-HIV drug development.

A Triazinone Derivative Inhibits HIV-1 Replication by Interfering with Reverse Transcriptase Activity

A novel HIV-1 inhibitor, 6-(tert-butyl)-4-phenyl-4-(trifluoromethyl)-1H,3H-1,3,5-triazin-2-one (compound 1), was identified from a compound library screened for the ability to inhibit HIV-1 replication. EC₅₀ values of compound 1 were found to range from 107.9 to 145.4 nm against primary HIV-1 clinical isolates. In in vitro assays, HIV-1 reverse transcriptase (RT) activity was inhibited by compound 1 with an EC₅₀ of 4.3 μm. An assay for resistance to compound 1 selected a variant of HIV-1 with a RT mutation (RT^L100I ); this frequently identified mutation confers mild resistance to non-nucleoside RT inhibitors (NNRTIs). A recombinant HIV-1 bearing RT^L100I exhibited a 41-fold greater resistance to compound 1 than the wild-type virus. Compound 1 was also effective against HIV-1 with RT^K103N , one of the major mutations that confers substantial resistance to NNRTIs. Computer-assisted docking simulations indicated that compound 1 binds to the RT NNRTI binding pocket in a manner similar to that of efavirenz; however, the putative compound 1 binding site is located further from RT^K103 than that of efavirenz. Compound 1 is a novel NNRTI with a unique drug-resistance profile.

Unprecedently Large-Scale Kinase Inhibitor Set Enabling the Accurate Prediction of Compound-Kinase Activities: A Way toward Selective Promiscuity by Design?

Drug discovery programs frequently target members of the human kinome and try to identify small molecule protein kinase inhibitors, primarily for cancer treatment, additional indications being increasingly investigated. One of the challenges is controlling the inhibitors degree of selectivity, assessed by in vitro profiling against panels of protein kinases. We manually extracted, compiled, and standardized such profiles published in the literature: we collected 356 908 data points corresponding to 482 protein kinases, 2106 inhibitors, and 661 patents. We then analyzed this data set in terms of kinome coverage, results reproducibility, popularity, and degree of selectivity of both kinases and inhibitors. We used the data set to create robust proteochemometric models capable of predicting kinase activity (the ligand-target space was modeled with an externally validated RMSE of 0.41 ± 0.02 log units and R02 0.74 ± 0.03), in order to account for missing or unreliable measurements. The influence on the prediction quality of parameters such as number of measurements, Murcko scaffold frequency or inhibitor type was assessed. Interpretation of the models enabled to highlight inhibitors and kinases properties correlated with higher affinities, and an analysis in the context of kinases crystal structures was performed. Overall, the models quality allows the accurate prediction of kinase-inhibitor activities and their structural interpretation, thus paving the way for the rational design of compounds with a targeted selectivity profile.

The recent progress in proteochemometric modelling: focusing on target descriptors, cross-term descriptors and application scope

As an extension of the conventional quantitative structure activity relationship models, proteochemometric (PCM) modelling is a computational method that can predict the bioactivity relations between multiple ligands and multiple targets. Traditional PCM modelling includes three essential elements: descriptors (including target descriptors, ligand descriptors and cross-term descriptors), bioactivity data and appropriate learning functions that link the descriptors to the bioactivity data. Since its appearance, PCM modelling has developed rapidly over the past decade by taking advantage of the progress of different descriptors and machine learning techniques, along with the increasing amounts of available bioactivity data. Specifically, the new emerging target descriptors and cross-term descriptors not only significantly increased the performance of PCM modelling but also expanded its application scope from traditional protein-ligand interaction to more abundant interactions, including protein-peptide, protein-DNA and even protein-protein interactions. In this review, target descriptors and cross-term descriptors, as well as the corresponding application scope, are intensively summarized. Additionally, we look forward to seeing PCM modelling extend into new application scopes, such as Target-Catalyst-Ligand systems, with the further development of descriptors, machine learning techniques and increasing amounts of available bioactivity data.

Modelling ligand selectivity of serine proteases using integrative proteochemometric approaches improves model performance and allows the multi-target dependent interpretation of features

Serine proteases, implicated in important physiological functions, have a high intra-family similarity, which leads to unwanted off-target effects of inhibitors with insufficient selectivity. However, the availability of sequence and structure data has now made it possible to develop approaches to design pharmacological agents that can discriminate successfully between their related binding sites. In this study, we have quantified the relationship between 12,625 distinct protease inhibitors and their bioactivity against 67 targets of the serine protease family (20,213 data points) in an integrative manner, using proteochemometric modelling (PCM). The benchmarking of 21 different target descriptors motivated the usage of specific binding pocket amino acid descriptors, which helped in the identification of active site residues and selective compound chemotypes affecting compound affinity and selectivity. PCM models performed better than alternative approaches (models trained using exclusively compound descriptors on all available data, QSAR) employed for comparison with R(2)/RMSE values of 0.64 ± 0.23/0.66 ± 0.20 vs. 0.35 ± 0.27/1.05 ± 0.27 log units, respectively. Moreover, the interpretation of the PCM model singled out various chemical substructures responsible for bioactivity and selectivity towards particular proteases (thrombin, trypsin and coagulation factor 10) in agreement with the literature. For instance, absence of a tertiary sulphonamide was identified to be responsible for decreased selective activity (by on average 0.27 ± 0.65 pChEMBL units) on FA10. Among the binding pocket residues, the amino acids (arginine, leucine and tyrosine) at positions 35, 39, 60, 93, 140 and 207 were observed as key contributing residues for selective affinity on these three targets.

Prediction of the potency of mammalian cyclooxygenase inhibitors with ensemble proteochemometric modeling

Cyclooxygenases (COX) are present in the body in two isoforms, namely: COX-1, constitutively expressed, and COX-2, induced in physiopathological conditions such as cancer or chronic inflammation. The inhibition of COX with non-steroideal anti-inflammatory drugs (NSAIDs) is the most widely used treatment for chronic inflammation despite the adverse effects associated to prolonged NSAIDs intake. Although selective COX-2 inhibition has been shown not to palliate all adverse effects (e.g. cardiotoxicity), there are still niche populations which can benefit from selective COX-2 inhibition. Thus, capitalizing on bioactivity data from both isoforms simultaneously would contribute to develop COX inhibitors with better safety profiles. We applied ensemble proteochemometric modeling (PCM) for the prediction of the potency of 3,228 distinct COX inhibitors on 11 mammalian cyclooxygenases. Ensemble PCM models ([Formula: see text], and RMSEtest = 0.71) outperformed models exclusively trained on compound ([Formula: see text], and RMSEtest = 1.09) or protein descriptors ([Formula: see text] and RMSEtest = 1.10) on the test set. Moreover, PCM predicted COX potency for 1,086 selective and non-selective COX inhibitors with [Formula: see text] and RMSEtest = 0.76. These values are in agreement with the maximum and minimum achievable [Formula: see text] and RMSEtest values of approximately 0.68 for both metrics. Confidence intervals for individual predictions were calculated from the standard deviation of the predictions from the individual models composing the ensembles. Finally, two substructure analysis pipelines singled out chemical substructures implicated in both potency and selectivity in agreement with the literature. Graphical AbstractPrediction of uncorrelated bioactivity profiles for mammalian COX inhibitors with Ensemble Proteochemometric Modeling.

Defective hydrophobic sliding mechanism and active site expansion in HIV-1 protease drug resistant variant Gly48Thr/Leu89Met: mechanisms for the loss of saquinavir binding potency

HIV drug resistance continues to emerge; consequently, there is an urgent need to develop next generation antiretroviral therapeutics.1 Here we report on the structural and kinetic effects of an HIV protease drug resistant variant with the double mutations Gly48Thr and Leu89Met (PR_G48T/L89M), without the stabilizing mutations Gln7Lys, Leu33Ile, and Leu63Ile. Kinetic analyses reveal that PR_G48T/L89M and PR_WT share nearly identical Michaelis–Menten parameters; however, PR_G48T/L89M exhibits weaker binding for IDV (41-fold), SQV (18-fold), APV (15-fold), and NFV (9-fold) relative to PR_WT. A 1.9 Å resolution crystal structure was solved for PR_G48T/L89M bound with saquinavir (PR_{G48T/L89M-SQV}) and compared to the crystal structure of PR_WT bound with saquinavir (PR_WT-SQV). PR_{G48T/L89M-SQV} has an enlarged active site resulting in the loss of a hydrogen bond in the S3 subsite from Gly48 to P3 of SQV, as well as less favorable hydrophobic packing interactions between P1 Phe of SQV and the S1 subsite. PR_{G48T/L89M-SQV} assumes a more open conformation relative to PR_WT-SQV, as illustrated by the downward displacement of the fulcrum and elbows and weaker interatomic flap interactions. We also show that the Leu89Met mutation disrupts the hydrophobic sliding mechanism by causing a redistribution of van der Waals interactions in the hydrophobic core in PR_{G48T/L89M-SQV}. Our mechanism for PR_{G48T/L89M-SQV} drug resistance proposes that a defective hydrophobic sliding mechanism results in modified conformational dynamics of the protease. As a consequence, the protease is unable to achieve a fully closed conformation that results in an expanded active site and weaker inhibitor binding.

Polypharmacology modelling using proteochemometrics (PCM): recent methodological developments, applications to target families, and future prospects

Proteochemometric (PCM) modelling is a computational method to model the bioactivity of multiple ligands against multiple related protein targets simultaneously.

HIV-1 diversity, transmission dynamics and primary drug resistance in Angola

Objectives

To assess HIV-1 diversity, transmission dynamics and prevalence of transmitted drug resistance (TDR) in Angola, five years after ART scale-up.

Methods

Population sequencing of the pol gene was performed on 139 plasma samples collected in 2009 from drug-naive HIV-1 infected individuals living in Luanda. HIV-1 subtypes were determined using phylogenetic analysis. Drug resistance mutations were identified using the Calibrated Population Resistance Tool (CPR). Transmission networks were determined using phylogenetic analysis of all Angolan sequences present in the databases. Evolutionary trends were determined by comparison with a similar survey performed in 2001.

Results

47.1% of the viruses were pure subtypes (all except B), 47.1% were recombinants and 5.8% were untypable. The prevalence of subtype A decreased significantly from 2001 to 2009 (40.0% to 10.8%, P = 0.0019) while the prevalence of unique recombinant forms (URFs) increased>2-fold (40.0% to 83.1%, P<0.0001). The most frequent URFs comprised untypable sequences with subtypes H (U/H, n = 7, 10.8%), A (U/A, n = 6, 9.2%) and G (G/U, n = 4, 6.2%). Newly identified U/H recombinants formed a highly supported monophyletic cluster suggesting a local and common origin. TDR mutation K103N was found in one (0.7%) patient (1.6% in 2001). Out of the 364 sequences sampled for transmission network analysis, 130 (35.7%) were part of a transmission network. Forty eight transmission clusters were identified; the majority (56.3%) comprised sequences sampled in 2008–2010 in Luanda which is consistent with a locally fuelled epidemic. Very low genetic distance was found in 27 transmission pairs sampled in the same year, suggesting recent transmission events.

Conclusions

Transmission of drug resistant strains was still negligible in Luanda in 2009, five years after the scale-up of ART. The dominance of small and recent transmission clusters and the emergence of new URFs are consistent with a rising HIV-1 epidemics mainly driven by heterosexual transmission.

Proteochemometric modeling in a Bayesian framework

Proteochemometrics (PCM) is an approach for bioactivity predictive modeling which models the relationship between protein and chemical information. Gaussian Processes (GP), based on Bayesian inference, provide the most objective estimation of the uncertainty of the predictions, thus permitting the evaluation of the applicability domain (AD) of the model. Furthermore, the experimental error on bioactivity measurements can be used as input for this probabilistic model.

In this study, we apply GP implemented with a panel of kernels on three various (and multispecies) PCM datasets. The first dataset consisted of information from 8 human and rat adenosine receptors with 10,999 small molecule ligands and their binding affinity. The second consisted of the catalytic activity of four dengue virus NS3 proteases on 56 small peptides. Finally, we have gathered bioactivity information of small molecule ligands on 91 aminergic GPCRs from 9 different species, leading to a dataset of 24,593 datapoints with a matrix completeness of only 2.43%.

GP models trained on these datasets are statistically sound, at the same level of statistical significance as Support Vector Machines (SVM), with R02 values on the external dataset ranging from 0.68 to 0.92, and RMSEP values close to the experimental error. Furthermore, the best GP models obtained with the normalized polynomial and radial kernels provide intervals of confidence for the predictions in agreement with the cumulative Gaussian distribution. GP models were also interpreted on the basis of individual targets and of ligand descriptors. In the dengue dataset, the model interpretation in terms of the amino-acid positions in the tetra-peptide ligands gave biologically meaningful results.

Towards predictive resistance models for agrochemicals by combining chemical and protein similarity via proteochemometric modelling

Resistance to pesticides is an increasing problem in agriculture. Despite practices such as phased use and cycling of ‘orthogonally resistant’ agents, resistance remains a major risk to national and global food security. To combat this problem, there is a need for both new approaches for pesticide design, as well as for novel chemical entities themselves. As summarized in this opinion article, a technique termed ‘proteochemometric modelling’ (PCM), from the field of chemoinformatics, could aid in the quantification and prediction of resistance that acts via point mutations in the target proteins of an agent. The technique combines information from both the chemical and biological domain to generate bioactivity models across large numbers of ligands as well as protein targets. PCM has previously been validated in prospective, experimental work in the medicinal chemistry area, and it draws on the growing amount of bioactivity information available in the public domain. Here, two potential applications of proteochemometric modelling to agrochemical data are described, based on previously published examples from the medicinal chemistry literature.

Chemical, target, and bioactive properties of allosteric modulation

Allosteric modulators are ligands for proteins that exert their effects via a different binding site than the natural (orthosteric) ligand site and hence form a conceptually distinct class of ligands for a target of interest. Here, the physicochemical and structural features of a large set of allosteric and non-allosteric ligands from the ChEMBL database of bioactive molecules are analyzed. In general allosteric modulators are relatively smaller, more lipophilic and more rigid compounds, though large differences exist between different targets and target classes. Furthermore, there are differences in the distribution of targets that bind these allosteric modulators. Allosteric modulators are over-represented in membrane receptors, ligand-gated ion channels and nuclear receptor targets, but are underrepresented in enzymes (primarily proteases and kinases). Moreover, allosteric modulators tend to bind to their targets with a slightly lower potency (5.96 log units versus 6.66 log units, p<0.01). However, this lower absolute affinity is compensated by their lower molecular weight and more lipophilic nature, leading to similar binding efficiency and surface efficiency indices. Subsequently a series of classifier models are trained, initially target class independent models followed by finer-grained target (architecture/functional class) based models using the target hierarchy of the ChEMBL database. Applications of these insights include the selection of likely allosteric modulators from existing compound collections, the design of novel chemical libraries biased towards allosteric regulators and the selection of targets potentially likely to yield allosteric modulators on screening. All data sets used in the paper are available for download.

The physicochemistry and topography of ligand binding sites is generally conserved amongst related proteins, however, comparisons of the pharmacology of related targets (and even the same target) are often confounded by the existence of multiple, distinct, binding sites within the same protein. Importantly, these multiple binding sites can have ‘druggability’ or selectivity properties, and can therefore offer attractive novel approaches to develop new therapeutic agents. In this paper, sets of known ligands binding to the same target are classified as being either allosteric (binding at a site that is non-competitive for a natural ligand/substrate) or non-allosteric (binding at the same site as a natural substrate), it is demonstrated that there are differences in the profiles of ligands discovered empirically against these sites. Finally predictive models are developed with several useful applications in drug discovery.

A polymorphism at position 400 in the connection subdomain of HIV-1 reverse transcriptase affects sensitivity to NNRTIs and RNaseH activity

Reverse transcriptase (RT) plays an essential role in HIV-1 replication, and inhibition of this enzyme is a key component of HIV-treatment. However, the use of RT inhibitors can lead to the emergence of drug-resistant variants. Until recently, most clinically relevant resistance mutations were found in the polymerase domain of RT. Lately, an increasing number of resistance mutations has been identified in the connection and RNaseH domain. To further explore the role of these domains we analyzed the complete RT sequence of HIV-1 subtype B patients failing therapy. Position A/T400 in the connection subdomain is polymorphic, but the proportion of T400 increases from 41% in naïve patients to 72% in patients failing therapy. Previous studies suggested a role for threonine in conferring resistance to nucleoside RT inhibitors. Here we report that T400 also mediates resistance to non-nucleoside RT inhibitors. The susceptibility to NVP and EFV was reduced 5-fold and 2-fold, respectively, in the wild-type subtype B NL4.3 background. We show that substitution A400T reduces the RNaseH activity. The changes in enzyme activity are remarkable given the distance to both the polymerase and RNaseH active sites. Molecular dynamics simulations were performed, which provide a novel atomistic mechanism for the reduction in RNaseH activity induced by T400. Substitution A400T was found to change the conformation of the RNaseH primer grip region. Formation of an additional hydrogen bond between residue T400 and E396 may play a role in this structural change. The slower degradation of the viral RNA genome may provide more time for dissociation of the bound NNRTI from the stalled RT-template/primer complex, after which reverse transcription can resume.

Benchmarking of protein descriptor sets in proteochemometric modeling (part 2): modeling performance of 13 amino acid descriptor sets

Background

While a large body of work exists on comparing and benchmarking descriptors of molecular structures, a similar comparison of protein descriptor sets is lacking. Hence, in the current work a total of 13 amino acid descriptor sets have been benchmarked with respect to their ability of establishing bioactivity models. The descriptor sets included in the study are Z-scales (3 variants), VHSE, T-scales, ST-scales, MS-WHIM, FASGAI, BLOSUM, a novel protein descriptor set (termed ProtFP (4 variants)), and in addition we created and benchmarked three pairs of descriptor combinations. Prediction performance was evaluated in seven structure-activity benchmarks which comprise Angiotensin Converting Enzyme (ACE) dipeptidic inhibitor data, and three proteochemometric data sets, namely (1) GPCR ligands modeled against a GPCR panel, (2) enzyme inhibitors (NNRTIs) with associated bioactivities against a set of HIV enzyme mutants, and (3) enzyme inhibitors (PIs) with associated bioactivities on a large set of HIV enzyme mutants.

Results

The amino acid descriptor sets compared here show similar performance (<0.1 log units RMSE difference and <0.1 difference in MCC), while errors for individual proteins were in some cases found to be larger than those resulting from descriptor set differences ( > 0.3 log units RMSE difference and >0.7 difference in MCC). Combining different descriptor sets generally leads to better modeling performance than utilizing individual sets. The best performers were Z-scales (3) combined with ProtFP (Feature), or Z-Scales (3) combined with an average Z-Scale value for each target, while ProtFP (PCA8), ST-Scales, and ProtFP (Feature) rank last.

Conclusions

While amino acid descriptor sets capture different aspects of amino acids their ability to be used for bioactivity modeling is still – on average – surprisingly similar. Still, combining sets describing complementary information consistently leads to small but consistent improvement in modeling performance (average MCC 0.01 better, average RMSE 0.01 log units lower). Finally, performance differences exist between the targets compared thereby underlining that choosing an appropriate descriptor set is of fundamental for bioactivity modeling, both from the ligand- as well as the protein side.

Benchmarking of protein descriptor sets in proteochemometric modeling (part 1): comparative study of 13 amino acid descriptor sets

Background

While a large body of work exists on comparing and benchmarking of descriptors of molecular structures, a similar comparison of protein descriptor sets is lacking. Hence, in the current work a total of 13 different protein descriptor sets have been compared with respect to their behavior in perceiving similarities between amino acids. The descriptor sets included in the study are Z-scales (3 variants), VHSE, T-scales, ST-scales, MS-WHIM, FASGAI and BLOSUM, and a novel protein descriptor set termed ProtFP (4 variants). We investigate to which extent descriptor sets show collinear as well as orthogonal behavior via principal component analysis (PCA).

Results

In describing amino acid similarities, MSWHIM, T-scales and ST-scales show related behavior, as do the VHSE, FASGAI, and ProtFP (PCA3) descriptor sets. Conversely, the ProtFP (PCA5), ProtFP (PCA8), Z-Scales (Binned), and BLOSUM descriptor sets show behavior that is distinct from one another as well as both of the clusters above. Generally, the use of more principal components (>3 per amino acid, per descriptor) leads to a significant differences in the way amino acids are described, despite that the later principal components capture less variation per component of the original input data.

Conclusion

In this work a comparison is provided of how similar (and differently) currently available amino acids descriptor sets behave when converting structure to property space. The results obtained enable molecular modelers to select suitable amino acid descriptor sets for structure-activity analyses, e.g. those showing complementary behavior.

Non-nucleoside reverse transcriptase inhibitor (NNRTI) cross-resistance: implications for preclinical evaluation of novel NNRTIs and clinical genotypic resistance testing

OBJECTIVES

The introduction of two new non-nucleoside reverse transcriptase inhibitors (NNRTIs) in the past 5 years and the identification of novel NNRTI-associated mutations have made it necessary to reassess the extent of phenotypic NNRTI cross-resistance.

METHODS

We analysed a dataset containing 1975, 1967, 519 and 187 genotype-phenotype correlations for nevirapine, efavirenz, etravirine and rilpivirine, respectively. We used linear regression to estimate the effects of RT mutations on susceptibility to each of these NNRTIs.

RESULTS

Sixteen mutations at 10 positions were significantly associated with the greatest contribution to reduced phenotypic susceptibility (≥10-fold) to one or more NNRTIs, including: 14 mutations at six positions for nevirapine (K101P, K103N/S, V106A/M, Y181C/I/V, Y188C/L and G190A/E/Q/S); 10 mutations at six positions for efavirenz (L100I, K101P, K103N, V106M, Y188C/L and G190A/E/Q/S); 5 mutations at four positions for etravirine (K101P, Y181I/V, G190E and F227C); and 6 mutations at five positions for rilpivirine (L100I, K101P, Y181I/V, G190E and F227C). G190E, a mutation that causes high-level nevirapine and efavirenz resistance, also markedly reduced susceptibility to etravirine and rilpivirine. K101H, E138G, V179F and M230L mutations, associated with reduced susceptibility to etravirine and rilpivirine, were also associated with reduced susceptibility to nevirapine and/or efavirenz.

CONCLUSIONS

The identification of novel cross-resistance patterns among approved NNRTIs illustrates the need for a systematic approach for testing novel NNRTIs against clinical virus isolates with major NNRTI-resistance mutations and for testing older NNRTIs against virus isolates with mutations identified during the evaluation of a novel NNRTI.

Inferring multi-target QSAR models with taxonomy-based multi-task learning

BACKGROUND

A plethora of studies indicate that the development of multi-target drugs is beneficial for complex diseases like cancer. Accurate QSAR models for each of the desired targets assist the optimization of a lead candidate by the prediction of affinity profiles. Often, the targets of a multi-target drug are sufficiently similar such that, in principle, knowledge can be transferred between the QSAR models to improve the model accuracy. In this study, we present two different multi-task algorithms from the field of transfer learning that can exploit the similarity between several targets to transfer knowledge between the target specific QSAR models.

RESULTS

We evaluated the two methods on simulated data and a data set of 112 human kinases assembled from the public database ChEMBL. The relatedness between the kinase targets was derived from the taxonomy of the humane kinome. The experiments show that multi-task learning increases the performance compared to training separate models on both types of data given a sufficient similarity between the tasks. On the kinase data, the best multi-task approach improved the mean squared error of the QSAR models of 58 kinase targets.

CONCLUSIONS

Multi-task learning is a valuable approach for inferring multi-target QSAR models for lead optimization. The application of multi-task learning is most beneficial if knowledge can be transferred from a similar task with a lot of in-domain knowledge to a task with little in-domain knowledge. Furthermore, the benefit increases with a decreasing overlap between the chemical space spanned by the tasks.

Structure-based identification of OATP1B1/3 inhibitors

Several recent studies show that inhibition of the hepatic transport proteins organic anion-transporting polypeptide 1B1 (OATP1B1) and 1B3 (OATP1B3) can result in clinically relevant drug-drug interactions (DDI). To avoid late-stage development drug failures due to OATP1B-mediated DDI, predictive in vitro and in silico methods should be implemented at an early stage of the drug candidate evaluation process. In the present study, we first developed a high-throughput in vitro transporter inhibition assay for the OATP1B subfamily. A total of 2000 compounds were tested as potential modulators of the uptake of the OATP1B substrate sodium fluorescein, in OATP1B1- or 1B3-transfected Chinese hamster ovary cells. At an equimolar substrate-inhibitor concentration of 10 µM, 212 and 139 molecules were identified as OATP1B1 and OATP1B3 inhibitors, respectively (minimum 50% inhibition). For 69 compounds, previously not identified as OATP1B inhibitors, concentration-dependent inhibition was also determined, yielding Ki values ranging from 0.06 to 6.5 µM. Based on these in vitro data, we subsequently developed a proteochemometrics-based in silico model, which predicted OATP1B inhibitors in the test group (20% of the dataset) with high specificity (86%) and sensitivity (78%). Moreover, several physicochemical compound properties and substructures related to OATP1B1/1B3 inhibition or inactivity were identified. Finally, model performance was prospectively verified with a set of 54 compounds not included in the original dataset. This validation indicated that 80 and 74% of the compounds were correctly classified for OATP1B1 and OATP1B3 inhibition, respectively.