Thermodynamic approaches for the prediction of oral drug absorption

The present study is a comparative study of three equations, namely the Clausius–Clapeyron, Van’t Hoff and Hildebrand (to calculate crystal–liquid fugacity ratio (CLFR) of drug compounds), to select the best model in predicting the intestinal absorption and develop a new classification system based on dose number (D _o) and CLFR. The required thermodynamic parameters [melting point, enthalpy of fusion (ΔH _m) and the differential molar heat capacity (?C _pm)] were experimentally obtained by differential scanning calorimetry. Pharmacokinetic data [the human intestinal absorption (F _a) and apparent permeability of Caco-2 (P _app _Caco-2)] and D _o were obtained from the literature. The highest value of CLFR was found for diclofenac with the value of 88.78, 87.29, and 87.84 mol% from Clausius–Clapeyron, Van’t Hoff, and Hildebrand approaches, respectively. The lowest CLFR value was seen for memantine with the value of 14.3 × 10^?17 and 26 × 10^?12 mol% from Van’t Hoff and Hildebrand equations, respectively. Statistical comparison with the Wilcoxon signed rank test showed that the CLFR values calculated by three equations are different. CLFR values of more than 1 mol% correspond to the complete intestinal absorption (F _a). There was a sigmoidal dependency between CLFR and F _a, similar to the dependency between P _app _Caco-2 and F _a. In these modeling, the excellent correlations were obtained in all three models as evidenced by a good coefficient of determination (r ² ) without a significant difference in the average absolute error. A new classification system from Hildebrand model based on D _o and CLFR was developed and was in agreement with the biopharmaceutics classification system (70.5%) and the biopharmaceutical drug disposition system (65.6%). This modeling approach can be a valuable tool for scientists as an alternative for intestinal permeability in the biopharmaceutical classification system to develop new oral drugs. The CLFR obtained from Hildebrand model is also more convenient than the Clausius Clapeyron model, because the former does not need to calculate ?C _pm (difficult step in calculating CLFR) for drug compounds. This new classification can help to develop the new drug product in industrial and academic research, without necessary in vivo experiments.     

Trypanosomal dUTPases as potential targets for drug design

Parasites of the Trypanosomatidae family are responsible for diseases that afflict several million people worldwide. Currently there is an urgent need for new drugs against these diseases and an approach to drug discovery is the study of biochemical and structural properties of a potential target and the subsequent design of specific compounds. Trypanosomatid genes coding for enzymes which distinctively hydrolyze dUTP have been isolated by genetic complementation in Escherichia coli mutants defective in dUTPase activity. An analysis of these sequences from Leishmania major and Trypanosoma cruzi showed that no significant similarity could be established with the family of known dUTPases and that the five consensus motifs were absent. However, limited similarity was identified for three motifs present in an enzyme related in function the dCTPase-dUTPase from T phages and 35 percent identity with a putative dUTPase identified in the eubacteria Campylobacter jejuni. T. cruzi and L. major dUTPases were highly similar and catalyzed in a specific fashion the hydrolysis of dUTP. A detailed kinetic study of both enzymes revealed that dUDP is also an efficient substrate of the enzyme while other nucleotides are poorly hydrolyzed. The enzyme is essential for viability in Leishmania and is up-regulated by inhibitors of dTMP synthesis. Thus, a new family of dUTPases might exist in certain organisms that bear no sequence or structure similarity with eukaryotic enzymes accomplishing the same function and that may constitute potential drug targets for the development of specific inhibitors.     

Learning size-adaptive molecular substructures for explainable drug–drug interaction prediction by substructure-aware graph neural network

Drug–drug interactions (DDIs) can trigger unexpected pharmacological effects on the body, and the causal mechanisms are often unknown. Graph neural networks (GNNs) have been developed to better understand DDIs. However, identifying key substructures that contribute most to the DDI prediction is a challenge for GNNs. In this study, we presented a substructure-aware graph neural network, a message passing neural network equipped with a novel substructure attention mechanism and a substructure–substructure interaction module (SSIM) for DDI prediction (SA-DDI). Specifically, the substructure attention was designed to capture size- and shape-adaptive substructures based on the chemical intuition that the sizes and shapes are often irregular for functional groups in molecules. DDIs are fundamentally caused by chemical substructure interactions. Thus, the SSIM was used to model the substructure–substructure interactions by highlighting important substructures while de-emphasizing the minor ones for DDI prediction. We evaluated our approach in two real-world datasets and compared the proposed method with the state-of-the-art DDI prediction models. The SA-DDI surpassed other approaches on the two datasets. Moreover, the visual interpretation results showed that the SA-DDI was sensitive to the structure information of drugs and was able to detect the key substructures for DDIs. These advantages demonstrated that the proposed method improved the generalization and interpretation capability of DDI prediction modeling.

SA-DDI is designed to learn size-adaptive molecular substructures for drug–drug interaction prediction and can provide explanations that are consistent with pharmacologists.     

Rational drug design and high-throughput techniques for RNA targets

RNA molecules are the only known molecules which possess the double property of being depository of genetic information, like DNA, and of displaying catalytic activities, like protein enzymes. RNA molecules intervene in all steps of gene expression and in many other biological activities. Like proteins, RNAs achieve those biological functions by adopting intricate three-dimensional folds and architectures. Further, as in protein sequences, RNA sequences contain signatures specific for three-dimensional motifs which participate in recognition and binding. In regulatory pathways, RNA molecules exist in equilibria between transient structures differentially stabilized by effectors such as proteins or cofactors. Therefore, RNA molecules display their potential as drug targets on different levels, namely in three-dimensional folds, in structural equilibria and in RNA-protein interfaces. Several examples will be described together with the already available techniques for combinatorial synthesis and high-throughput screening of potential drug and target RNA molecules.     

BonDNet: a graph neural network for the prediction of bond dissociation energies for charged molecules

A broad collection of technologies, including e.g. drug metabolism, biofuel combustion, photochemical decontamination of water, and interfacial passivation in energy production/storage systems rely on chemical processes that involve bond-breaking molecular reactions. In this context, a fundamental thermodynamic property of interest is the bond dissociation energy (BDE) which measures the strength of a chemical bond. Fast and accurate prediction of BDEs for arbitrary molecules would lay the groundwork for data-driven projections of complex reaction cascades and hence a deeper understanding of these critical chemical processes and, ultimately, how to reverse design them. In this paper, we propose a chemically inspired graph neural network machine learning model, BonDNet, for the rapid and accurate prediction of BDEs. BonDNet maps the difference between the molecular representations of the reactants and products to the reaction BDE. Because of the use of this difference representation and the introduction of global features, including molecular charge, it is the first machine learning model capable of predicting both homolytic and heterolytic BDEs for molecules of any charge. To test the model, we have constructed a dataset of both homolytic and heterolytic BDEs for neutral and charged (−1 and +1) molecules. BonDNet achieves a mean absolute error (MAE) of 0.022 eV for unseen test data, significantly below chemical accuracy (0.043 eV). Besides the ability to handle complex bond dissociation reactions that no previous model could consider, BonDNet distinguishes itself even in only predicting homolytic BDEs for neutral molecules; it achieves an MAE of 0.020 eV on the PubChem BDE dataset, a 20% improvement over the previous best performing model. We gain additional insight into the model''s predictions by analyzing the patterns in the features representing the molecules and the bond dissociation reactions, which are qualitatively consistent with chemical rules and intuition. BonDNet is just one application of our general approach to representing and learning chemical reactivity, and it could be easily extended to the prediction of other reaction properties in the future.

Prediction of bond dissociation energies for charged molecules with a graph neural network enabled by global molecular features and reaction difference features between products and reactants.     

Cytokine antibody arrays: a promising tool to identify molecular targets for drug discovery

Cytokines play important roles in normal cell functions and changes in cytokines have been implicated in many diseases. Recent efforts have focused on developing cytokine antibody arrays. These arrays allow investigators to simultaneously detect multiple cytokines in qualitative and quantitative ways. Cytokine antibody array systems feature high sensitivity, specificity and throughput. This novel technology opens up an expanding spectrum of applications in drug discovery, including target discovery, target validation, screening for lead compounds, compound optimization and clinical trials.     

In silico prediction of human genes as potential targets for rice miRNAs

BackgroundExogenous microRNAs (miRNAs) enter the human body through food, and their effects on metabolic processes can be considerable. It is important to determine which miRNAs from plants affect the expression of human genes and the extent of their influence.MethodThe binding sites of 738Oryza sativa miRNAs (osa-miRNAs) that interact with 17 508 mRNAs of human genes were determined using the MirTarget program.ResultThe characteristics of the binding of 46 single osa-miRNAs to 86 mRNAs of human genes with a value of free energy (ΔG) interaction equal 94%–100% from maximum ΔG were established. The findings showed that osa-miR2102-5p, osa-miR5075-3p, osa-miR2097-5p, osa-miR2919 targeted the largest number of genes at 38, 36, 23, 19 sites, respectively. mRNAs of 86 human genes were identified as targets for 93 osa-miRNAs of all family osa-miRNAs with ΔG values equal 94%–98% from maximum ΔG. Each miRNA of the osa-miR156-5p, osa-miR164-5p, osa-miR168-5p, osa-miR395-3p, osa-miR396-3p, osa-miR396-5p, osa-miR444-3p, osa-miR529-3p, osa-miR1846-3p, osa-miR2907-3p families had binding sites in mRNAs of several human target genes. The binding sites of osa-miRNAs in mRNAs of the target genes for each family of osa-miRNAs were conserved when compared to flanking nucleotide sequences.ConclusionTarget mRNA human genes of osa-miRNAs are also candidate genes of cancer, cardiovascular and neurodegenerative diseases.     

The DNA three-way junction as a mould for tripartite chromophore assembly

The DNA three-way junction serves as a scaffold for the molecular organization of non-nucleosidic alkynylpyrene and perylenediimide chromophores located at the branch point of the structure. Depending on the composition of the tripartite assembly, the constructs possess distinct spectroscopic properties, ranging from monomer or excimer fluorescence to completely quenched tripartite aggregates.     

In silico prediction of drug toxicity

It is essential, in order to minimise expensive drug failures due to toxicity being found in late development or even in clinical trials, to determine potential toxicity problems as early as possible. In view of the large libraries of compounds now being handled by combinatorial chemistry and high-throughput screening, identification of putative toxicity is advisable even before synthesis. Thus the use of predictive toxicology is called for. A number of in silico approaches to toxicity prediction are discussed. Quantitative structure-activity relationships (QSARs), relating mostly to specific chemical classes, have long been used for this purpose, and exist for a wide range of toxicity endpoints. However, QSARs also exist for the prediction of toxicity of very diverse libraries, although often such QSARs are of the classification type; that is, they predict simply whether or not a compound is toxic, and do not give an indication of the level of toxicity. Examples are given of all of these. A number of expert systems are available for toxicity prediction, most of them covering a range of toxicity endpoints. Those discussed include TOPKAT, CASE, DEREK, HazardExpert, OncoLogic and COMPACT. Comparative tests of the ability of these systems to predict carcinogenicity show that improvement is still needed. The consensus approach is recommended, whereby the results from several prediction systems are pooled. It is simply amazing that we can formulate any kind of QSAR. The (desired activity) is only the starting point. The truly formidable problem is that of toxicity, especially the difficult long-term toxicities resulting from chronic usage'. (Hansch & Leo [1])     

In-line prediction of drug release profiles for pH-sensitive coated pellets

A new method for the prediction of the drug release profiles during a running pellet coating process from in-line near infrared (NIR) measurements has been developed. The NIR spectra were acquired during a manufacturing process through an immersion probe. These spectra reflect the coating thickness that is inherently connected with the drug release. Pellets sampled at nine process time points from thirteen designed laboratory-scale coating batches were subjected to the dissolution testing. In the case of the pH-sensitive Acryl-EZE coating the drug release kinetics for the acidic medium has a sigmoid form with a pronounced induction period that tends to grow along with the coating thickness. In this work the autocatalytic model adopted from the chemical kinetics has been successfully applied to describe the drug release. A generalized interpretation of the kinetic constants in terms of the process and product parameters has been suggested. A combination of the kinetic model with the multivariate Partial Least Squares (PLS) regression enabled prediction of the release profiles from the process NIR data. The method can be used to monitor the final pellet quality in the course of a coating process.     

A new method for side-chain conformation prediction using a Hopfield network and reproduced rotamers

We present a new side-chain prediction method based on energy minimization using a Hopfield network, focusing on the buried residues of proteins. In this method, the network is composed of automata assigned to each rotamer to restrict side-chain conformational space. We reproduced a rotamer library that enabled us to more widely cover the space for side-chain conformations than those previously produced. The accuracy of the side-chain modeling was estimated by three standards: root mean square deviations (rmsds) between the modeled and the crystal structures, the percentages of correctly predicted side-chain torsion angles, and the percentages of correctly predicted hydrogen bonds. The average rmsd for buried side chains of 21 proteins was 1.10 Å. The value was almost always improved relative to the previous works. The percentage of side-chain X₁ angles for buried residues was 87.3%. By considering the hydrogen bond energy, the average percentage of correctly predicted hydrogen bonds rose from 33% without hydrogen bond energy to 52% with the bond energy. We applied this method to homology modeling, where the protein backbone used to predict side-chain conformations deviates from the correct conformation, and could predict side-chain conformations as correctly as those using the correct backbones. © 1996 by John Wiley & Sons, Inc.     

Design and characterization of a functional library for NMR screening against novel protein targets

In the past few years, NMR has been extensively utilized as a screening tool for drug discovery using various types of compound libraries. The designs of NMR specific chemical libraries that utilize a fragment-based approach based on drug-like characteristics have been previously reported. In this article, a new type of compound library will be described that focuses on aiding in the functional annotation of novel proteins that have been identified from various ongoing genomics efforts. The NMR functional chemical library is comprised of small molecules with known biological activity such as: co-factors, inhibitors, metabolites and substrates. This functional library was developed through an extensive manual effort of mining several databases based on known ligand interactions with protein systems. In order to increase the efficiency of screening the NMR functional library, the compounds are screened as mixtures of 3-4 compounds that avoids the need to deconvolute positive hits by maintaining a unique NMR resonance and function for each compound in the mixture. The functional library has been used in the identification of general biological function of hypothetical proteins identified from the Protein Structure Initiative.     

Multicopy suppressors for novel antibacterial compounds reveal targets and drug efflux susceptibility

Gene dosage has frequently been exploited to select for genetic interactions between a particular mutant and clones from a random genomic library at high copy. We report here the first use of multicopy suppression as a forward genetic method to determine cellular targets and potential resistance mechanisms for novel antibacterial compounds identified through high-throughput screening. A screen of 8640 small molecules for growth inhibition of a hyperpermeable strain of Escherichia coli led to the identification of 49 leads for suppressor selection from clones harboring an E. coli genomic library. The majority of suppressors were found to encode the multidrug efflux pump AcrB, indicating that those compounds were substrates for efflux. Two leads, which produced clones containing the gene folA, encoding dihydrofolate reductase (DHFR), proved to target DHFR in vivo and were competitive inhibitors in vitro.     

Design and training of a neural network for predicting the solvent accessibility of proteins

A feed-forward neural network has been developed to predict the solvent accessibility/accessible surface area (ASA) of proteins using improved design and training methods. Several network issues ranging from the coding of ASA states to the problem of local minima of learning curve, have been addressed. Successful new approaches to overcome these problems are presented. Set of trained network weights for each ASA threshold is provided. It has been established that the prediction accuracy results with neural network are better than other reported results of ASA prediction, despite a high test to training data ratio.     

Toward designing drug-like libraries: a novel computational approach for prediction of drug feasibility of compounds

Prediction of the degree of drug-like character in small molecules is of great industrial interest. The major barrier, however, is the lack of a definition for drug-like character. We used the concept of the multilevel chemical compatibility (MLCC) between a compound and a drug library as a measure of the drug-like character of a compound. The rationale is that the local chemical environment of each atom or group of atoms in a compound largely contributes to the stability, toxicity, and metabolism in vivo. A systematic comparison of the local environments within a compound and those within the existing drugs provides a basis for determining whether and how much a compound is drug-like. We applied the MLCC calculations to four test sets: top selling drugs, compounds under biological testing prior to the preclinical test, anticancer drugs, and compounds known to have poor drug-like character. The following conclusions were obtained: (1) A convergent number of unique local structure types were found in the analysis of the library of the existing drugs. It suggests that the current drug library contains about 80% of all the viable types; therefore, discovery of a drug with new local structures is only an event of relatively small probability. (2) The method is highly selective in discerning drug-like compounds: most of the top drugs are predicted to be drug-like, about one-quarter of the biological testing compounds are drug-like, and about one-fifth of the anticancer drugs are drug-like. (3) The method also correctly predicted that none of the known problematic compounds are drug-like. (4) The method is fast enough for computational screening of virtual combinatorial chemistry libraries and databases of available compounds.