An alternative biomimetic tool – Dual hydrophilic/reversed-phase interaction mode

Author index

Multidimensional compound optimization is a new paradigm in the drug discovery process, yielding efficiencies during early stages and reducing attrition in the later stages of drug development. The success of this strategy relies heavily on understanding this multidimensional data and extracting useful information from it. This paper demonstrates how principled visualization algorithms can be used to understand and explore a large data set created in the early stages of drug discovery. The experiments presented are performed on a real-world data set comprising biological activity data and some whole-molecular physicochemical properties. Data visualization is a popular way of presenting complex data in a simpler form. We have applied powerful principled visualization methods, such as generative topographic mapping (GTM) and hierarchical GTM (HGTM), to help the domain experts (screening scientists, chemists, biologists, etc.) understand and draw meaningful decisions. We also benchmark these principled methods against relatively better known visualization approaches, principal component analysis (PCA), Sammon's mapping, and self-organizing maps (SOMs), to demonstrate their enhanced power to help the user visualize the large multidimensional data sets one has to deal with during the early stages of the drug discovery process. The results reported clearly show that the GTM and HGTM algorithms allow the user to cluster active compounds for different targets and understand them better than the benchmarks. An interactive software tool supporting these visualization algorithms was provided to the domain experts. The tool facilitates the domain experts by exploration of the projection obtained from the visualization algorithms providing facilities such as parallel coordinate plots, magnification factors, directional curvatures, and integration with industry standard software.

The field of drug development and discovery encounters a number of challenges in identifying effective therapeutics to treat central nervous system (CNS) diseases. Experimental methods such as pharmacokinetic/pharmacodynamic (PK/PD) modeling and behavioral testing have been the conventional means to assess drug efficacy. Here we introduce functional Magnetic Resonance Imaging (fMRI) as a complimentary technique that can be implemented to assess the effectiveness of a therapeutic in CNS disease. FMRI has the unique ability to characterize the de novo drug effect on specific CNS targets, as is done in Positron Emission Tomography, as well as determine how neuronal substrates or networks are influenced by the therapeutic of interest during sensory stimulation or cognitive and motor tasks. Furthermore, fMRI measures can easily be related to the results obtained from conventional standards, such as PK/PD modeling. FMRI is believed to be a promising experimental method that can assist in defining drug effect in early stages of drug development and discovery; and thus, improve the go-no-go decision-making process of newly identified drugs to treat CNS disease.

ABSTRACTPharmacokinetic studies play an important role in all stages of drug discovery and development. Recent advancements in the tools for discovery and optimization of therapeutic proteins have created an abundance of candidates that may fulfill target product profile criteria. Implementing a set of in silico, small scale in vitro and in vivo tools can help to identify a clinical lead molecule with promising properties at the early stages of drug discovery, thus reducing the labor and cost in advancing multiple candidates toward clinical development. In this review, we describe tools that should be considered during drug discovery, and discuss approaches that could be included in the pharmacokinetic screening part of the lead candidate generation process to de-risk unexpected pharmacokinetic behaviors of Fc-based therapeutic proteins, with an emphasis on monoclonal antibodies.

A goal of preclinical discovery is the identification of drug candidates suitable for clinical testing. Successful integration of in vitro and in vivo experimental data sets can afford projections of human dose regimens anticipated to be safe and therapeutically beneficial. While in vitro experiments guide new chemical syntheses and are essential to understanding drug action and disposition, in vivo characterizations provide unique insight into complex biological systems that control concentrations at the site of action and pharmacologic response. Pharmacokinetic and pharmacodynamic (PK/PD) concepts underlying drug disposition and response provide a quantitative framework with which to identify potential clinical candidates. To improve throughput in earlier stages of drug discovery, in vivo pharmacokinetic study designs such as cassette dosing and sparse sampling schemes have been utilized. In later stages of discovery, pharmacokinetic studies using chemical inhibitors or surgical and genetic animal models are used to characterize the underlying determinants of drug disposition. In a complimentary fashion, modeling of in vivo pharmacodynamic effects may quantitatively link biomarkers to pharmacological response, validate in vitro to in vivo correlations and underwrite predictions of efficacious exposure targets. When applied to in vivo discovery data, PK/PD models have aided in understanding mechanisms of pharmacological response such as receptor theory in the central nervous system and cell turnover concepts in infectious disease and oncology. This review considers the role of in vivo testing toward understanding the pharmacokinetic and pharmacodynamic attributes of lead candidates in drug discovery.

Liver toxicity is a major adverse drug reaction that accounts for drug failure in clinical trials and withdrawal from the market. Therefore, predicting potential liver toxicity at an early stage in drug discovery is crucial to reduce costs and the potential for drug failure. However, current in vivo animal toxicity testing is very expensive and time consuming. As an alternative approach, various machine learning models have been developed to predict potential liver toxicity in humans. This chapter reviews current advances in the development and application of machine learning models for prediction of potential liver toxicity in humans and discusses possible improvements to liver toxicity prediction.

Requirements for predicting human pharmacokinetics in drug discovery are increasing. Developing different methods of human pharmacokinetic prediction will facilitate lead optimization, candidate nomination, and dosing regimens before clinical trials at various early drug discovery stages. To develop and validate generic methods of human pharmacokinetic prediction to meet the requirements in early drug discovery. The physiologically based pharmacokinetic (PBPK) model implemented in Gastroplus™ was used for human pharmacokinetic predictions. The absorption, distribution, metabolism, and excretion properties of drugs in humans predicted from molecular structure and extrapolated from tested preclinical data were used as inputs in the PBPK model. The approaches were validated by comparison of the predicted pharmacokinetic parameters with actual pharmacokinetic parameters of 15 marketed small-molecule compounds approved by the US Food and Drug Administration. Based on the validation and reported approaches, we proposed a strategy for human pharmacokinetic prediction at different drug discovery stages. Obvious underestimation of exposure (< 1/3 of actual exposure) was not observed using in silico prediction as inputs, which may reduce the probability of missing the potential compounds with predicted false low exposure. The simulated human pharmacokinetic results using tested data as inputs were superior to those obtained via in silico prediction. Both methods similarly predicted the multiphasic shape of pharmacokinetic profiles. These generic PBPK approaches of full in silico prediction or perdition using a combination of tested in vivo and in vitro data were validated and proved useful for human pharmacokinetic predictions.

The study presents 'G4-QuadScreen', a user-friendly computational tool for identifying MTDLs against G4s. Also, it offers a few hit MTDLs based on in silico and in vitro approaches. Multi-tasking QSAR models were developed using linear discriminant analysis and random forest machine learning techniques for predicting the responses of interest (G4 interaction, G4 stabilization, G4 selectivity, and cytotoxicity) considering the variations in the experimental conditions (e.g., G4 sequences, endpoints, cell lines, buffers, and assays). A virtual screening with G4-QuadScreen and molecular docking using YASARA (AutoDock-Vina) was performed. G4 activities were confirmed via FRET melting, FID, and cell viability assays. Validation metrics demonstrated the high discriminatory power and robustness of the models (the accuracy of all models is ~>90% for the training sets and ~>80% for the external sets). The experimental evaluations showed that ten screened MTDLs have the capacity to selectively stabilize multiple G4s. Three screened MTDLs induced a strong inhibitory effect on various human cancer cell lines. This pioneering computational study serves a tool to accelerate the search for new leads against G4s, reducing false positive outcomes in the early stages of drug discovery. The G4-QuadScreen tool is accessible on the ChemoPredictionSuite website.

State‑of‑the‑art medical studies proved that predicting CYP450 enzyme inhibitors is beneficial in the early stage of drug discovery. However, accurate machine learning-based (ML) in silico methods for predicting CYP450 inhibitors remains challenging. Here, we introduce GTransCYPs, an improved graph neural network (GNN) with a transformer mechanism for predicting CYP450 inhibitors. This model significantly enhances the discrimination between inhibitors and non-inhibitors for five major CYP450 isozymes: 1A2, 2C9, 2C19, 2D6, and 3A4. GTransCYPs learns information patterns from molecular graphs by aggregating node and edge representations using a transformer. The GTransCYPs model utilizes transformer convolution layers to process features, followed by a global attention-pooling technique to synthesize the graph-level information. This information is then fed through successive linear layers for final output generation. Experimental results demonstrate that the GTransCYPs model achieved high performance, outperforming other state-of-the-art methods in CYP450 prediction.Scientific contributionThe prediction of CYP450 inhibition via computational techniques utilizing biological information has emerged as a cost-effective and highly efficient approach. Here, we presented a deep learning (DL) architecture based on GNN with transformer mechanism and attention pooling (GTransCYPs) to predict CYP450 inhibitors. Four GTransCYPs of different pooling technique were tested on an experimental tasks on the CYP450 prediction problem for the first time. Graph transformer with attention pooling algorithm achieved the best performances. Comparative and ablation experiments provide evidence of the efficacy of our proposed method in predicting CYP450 inhibitors. The source code is publicly available at https://github.com/zonwoo/GTransCYPs.

Cell-based high-throughput screening for the evaluation of reactive metabolite formation potential

Poor metabolic stability leads to drug development failure. Therefore, it is essential to evaluate the metabolic stability of small compounds for successful drug discovery and development. However, evaluating metabolic stability in vitro and in vivo is expensive, time-consuming and laborious. In addition, only a few free software programs are available for metabolic stability data and prediction. Therefore, in this study, we aimed to develop a prediction model that predicts the metabolic stability of small compounds. We developed a computational model, PredMS, which predicts the metabolic stability of small compounds as stable or unstable in human liver microsomes. PredMS is based on a random forest model using an in-house database of metabolic stability data of 1917 compounds. To validate the prediction performance of PredMS, we generated external test data of 61 compounds. PredMS achieved an accuracy of 0.74, Matthew's correlation coefficient of 0.48, sensitivity of 0.70, specificity of 0.86, positive predictive value of 0.94 and negative predictive value of 0.46 on the external test dataset. PredMS will be a useful tool to predict the metabolic stability of small compounds in the early stages of drug discovery and development. The source code for PredMS is available at https://bitbucket.org/krictai/predms, and the PredMS web server is available at https://predms.netlify.app. Supplementary data are available at Bioinformatics online.

PSY1-2 - Key Points for Simultaneous Development and Approval of Medicine and Companion Diagnostics

Electrochemistry in combination with mass spectrometry is emerging as a versatile analytical technique in the imitation of oxidative drug metabolism during the early stages of drug discovery and development. Here, we present electrochemical O-dealkylation of phenacetin to acetaminophen by square-wave potential pulses consisting of consecutive sub-second oxidation and reduction steps. This O-dealkylation could not be achieved by oxidation at constant potential or longer potential pulses because of the fast hydrolysis of the reactive intermediates. Electrochemical conversion by square-wave potential pulses can thus widen the scope of electrochemical synthesis of metabolites and imitation of in vivo drug metabolism.

Aim: In the early stages of drug discovery, various experimental and computational methods are used to measure the specificity of small molecules against a target protein. The selectivity of small molecules remains a challenge leading to off-target side effects. Methods: We have developed a multitask deep learning model for predicting the selectivity on closely related homologs of the target protein. The model has been tested on the Janus-activated kinaseand dopamine receptor families of proteins. Results & conclusion: The feature-based representation (extended connectivity fingerprint 4) with Extreme Gradient Boosting performed better when compared with deep neural network models in most of the evaluation metrics. Both the Extreme Gradient Boostingand deep neural networkmodels outperformed the graph-based models. Furthermore, to decipher the model decision on selectivity, the important fragments associated with each homologous protein were identified.

In Vitro absorption and secretory quotients: Practical criteria derived from a study of 331 compounds to assess for the impact of P-glycoprotein-mediated efflux on drug candidates