The AI-driven Drug Design (AIDD) platform: an interactive multi-parameter optimization system integrating molecular evolution with physiologically based pharmacokinetic simulations.

Computer-aided drug design has advanced rapidly in recent years, and multiple instances of in silico designed molecules advancing to the clinic have demonstrated the contribution of this field to medicine. Properly designed and implemented platforms can drastically reduce drug development timelines and costs. While such efforts were initially focused primarily on target affinity/activity, it is now appreciated that other parameters are equally important in the successful development of a drug and its progression to the clinic, including pharmacokinetic properties as well as absorption, distribution, metabolic, excretion and toxicological (ADMET) properties. In the last decade, several programs have been developed that incorporate these properties into the drug design and optimization process and to varying degrees, allowing for multi-parameter optimization. Here, we introduce the Artificial Intelligence-driven Drug Design (AIDD) platform, which automates the drug design process by integrating high-throughput physiologically-based pharmacokinetic simulations (powered by GastroPlus) and ADMET predictions (powered by ADMET Predictor) with an advanced evolutionary algorithm that is quite different than current generative models. AIDD uses these and other estimates in iteratively performing multi-objective optimizations to produce novel molecules that are active and lead-like. Here we describe the AIDD workflow and details of the methodologies involved therein. We use a dataset of triazolopyrimidine inhibitors of the dihydroorotate dehydrogenase from Plasmodium falciparum to illustrate how AIDD generates novel sets of molecules.

pK50─A Rigorous Indicator of Individual Functional Group Acidity/Basicity in Multiprotic Compounds.

In this work, we show that the apparent pKa measured by standard titration experiments is an insufficient measure of acidity or basicity of organic functional groups in multiprotic compounds─a frequent aspect of lead optimization in pharmaceutical research. We show that the use of the apparent pKa in this context may result in costly mistakes. To properly represent the group's true acidity/basicity, we propose pK50─a single-proton midpoint measure derived from a statistical thermodynamics treatment of multiprotic ionization. We show that pK50, which may be directly measured in specialized NMR titration experiments, is superior in tracking the functional group's acidity/basicity across congeneric series of related compounds and converges to the well familiar ionization constant in the monoprotic case.

Using Cheminformatics in Drug Discovery.

This chapter illustrates how cheminformatics can be applied to designing novel compounds that are active at the primary target and have good predicted ADMET properties. Examples of various cheminformatics techniques are illustrated in the process of designing inhibitors that inhibit both cyclooxygenase isoforms but are more potent toward COX-2. The first step in the process is to create a knowledge database of cyclooxygenase inhibitors in the public domain. This data was analyzed to find activity cliffs - small structural changes that result in drastic changes in potency. Additional cyclooxygenase potency and selectivity trends were obtained using matched molecular pair analysis. QSAR models were then developed to predict cyclooxygenase potency and selectivity. Next, computational algorithms were used to generate novel scaffolds starting from known cyclooxygenase inhibitors. Nine virtual libraries containing 240 compounds each were constructed. Predictions from the cyclooxygenase QSAR models were used to eliminate molecules with undesirable potency or selectivity. Additionally, the compounds were screened in silico for undesirable ADMET properties, e.g., low solubility, permeability, metabolic stability, or high toxicity, using a liability scoring system known as ADMET Risk™. Eight synthetic candidates were identified from this process after incorporating knowledge gained from activity cliff analysis. Four of the compounds were synthesized and tested to measure their COX-1 and COX-2 IC(50) values as well as several ADME properties. The best compound, SLP0020, had a COX-1 IC(50) of 770 nM and COX-2 IC(50) of 130 nM.

Tales from the war on error: the art and science of curating QSAR data.

Curating the data underlying quantitative structure-activity relationship models is a never-ending struggle. Some curation can now be automated but much cannot, especially where data as complex as those pertaining to molecular absorption, distribution, metabolism, excretion, and toxicity are concerned (vide infra). The authors discuss some particularly challenging problem areas in terms of specific examples involving experimental context, incompleteness of data, confusion of units, problematic nomenclature, tautomerism, and misapplication of automated structure recognition tools.

Lions and tigers and bears, oh my! Three barriers to progress in computer-aided molecular design.

The computational chemistry and cheminformatics community faces many challenges to advancing the state of the art. We discuss three of those challenges here: accurately estimating the contribution of entropy to ligand binding; reliably estimating the uncertainties in model predictions for new molecules; and being able to effectively curate the ever-expanding literature and commercial databases needed to build new models.

Evidence-based medicine: how to translate research into patient care.

The purpose of this article is to review the history and development of evidence-based medicine, to provide a basic outline of its application to clinical care, and to discuss its pros and cons. This article can be used as a tool in podiatric medicine and surgery to ensure that current best evidence, clinical intuition, and patient preferences inform and guide our medical decision making.

Modeling ADMET.

Drug discovery and development is a costly and time-consuming endeavor (Calcoen et al. Nat Rev Drug Discov 14(3):161-162, 2015; The truly staggering cost of inventing new drugs. Forbes. http://www.forbes.com/sites/matthewherper/2012/02/10/the-truly-staggering-cost-of-inventing-new-drugs/, 2012; Scannell et al. Nat Rev Drug Discov 11(3):191-200, 2012). Over the last two decades, computational tools and in silico models to predict ADMET (Adsorption, Distribution, Metabolism, Excretion, and Toxicity) profiles of molecules have been incorporated into the drug discovery process mainly in an effort to avoid late-stage failures due to poor pharmacokinetics and toxicity. It is now widely recognized that ADMET issues should be addressed as early as possible in drug discovery. Here, we describe in detail how ADMET models can be developed and applied using a commercially available package, ADMET Predictor™ 7.2 (ADMET Predictor v7.2. Simulations Plus, Inc., Lancaster, CA, USA).

LigScore: a novel scoring function for predicting binding affinities.

We present two new empirical scoring functions, LigScore1 and LigScore2, that attempt to accurately predict the binding affinity between ligand molecules and their protein receptors. The LigScore functions consist of three distinct terms that describe the van der Waals interaction, the polar attraction between the ligand and protein, and the desolvation penalty attributed to the binding of the polar ligand atoms to the protein and vice versa. Utilizing a regression approach on a data set of 118 protein-ligand complexes we have obtained a linear equation, LigScore2, using these three descriptors. LigScore2 has good predictability with regard to experimental pKi values yielding a correlation coefficient, r2), of 0.75 and a standard deviation of 1.04 over the training data set, which consists of a diverse set of proteins that span more than seven protein families.

Design of focused and restrained subsets from extremely large virtual libraries.

With the current and ever-growing offering of reagents along with the vast palette of organic reactions, virtual libraries accessible to combinatorial chemists can reach sizes of billions of compounds or more. Extracting practical size subsets for experimentation has remained an essential step in the design of combinatorial libraries. A typical approach to computational library design involves enumeration of structures and properties for the entire virtual library, which may be unpractical for such large libraries. This study describes a new approach termed as on the fly optimization (OTFO) where descriptors are computed as needed within the subset optimization cycle and without intermediate enumeration of structures. Results reported herein highlight the advantages of coupling an ultra-fast descriptor calculation engine to subset optimization capabilities. We also show that enumeration of properties for the entire virtual library may not only be unpractical but also wasteful. Successful design of focused and restrained subsets can be achieved while sampling only a small fraction of the virtual library. We also investigate the stability of the method and compare results obtained from simulated annealing (SA) and genetic algorithms (GA).

Using beta binomials to estimate classification uncertainty for ensemble models.

BACKGROUND: Quantitative structure-activity (QSAR) models have enormous potential for reducing drug discovery and development costs as well as the need for animal testing. Great strides have been made in estimating their overall reliability, but to fully realize that potential, researchers and regulators need to know how confident they can be in individual predictions. RESULTS: Submodels in an ensemble model which have been trained on different subsets of a shared training pool represent multiple samples of the model space, and the degree of agreement among them contains information on the reliability of ensemble predictions. For artificial neural network ensembles (ANNEs) using two different methods for determining ensemble classification - one using vote tallies and the other averaging individual network outputs - we have found that the distribution of predictions across positive vote tallies can be reasonably well-modeled as a beta binomial distribution, as can the distribution of errors. Together, these two distributions can be used to estimate the probability that a given predictive classification will be in error. Large data sets comprised of logP, Ames mutagenicity, and CYP2D6 inhibition data are used to illustrate and validate the method. The distributions of predictions and errors for the training pool accurately predicted the distribution of predictions and errors for large external validation sets, even when the number of positive and negative examples in the training pool were not balanced. Moreover, the likelihood of a given compound being prospectively misclassified as a function of the degree of consensus between networks in the ensemble could in most cases be estimated accurately from the fitted beta binomial distributions for the training pool. CONCLUSIONS: Confidence in an individual predictive classification by an ensemble model can be accurately assessed by examining the distributions of predictions and errors as a function of the degree of agreement among the constituent submodels. Further, ensemble uncertainty estimation can often be improved by adjusting the voting or classification threshold based on the parameters of the error distribution. Finally, the profiles for models whose predictive uncertainty estimates are not reliable provide clues to that effect without the need for comparison to an external test set.