Robert Daniel Clark

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.

Quantitative structure-activity relationship (QSAR) models are powerful in silico tools for predicting the mutagenicity of unstable compounds, impurities and metabolites that are difficult to examine using the Ames test. Ideally, Ames/QSAR models for regulatory use should demonstrate high sensitivity, low false-negative rate and wide coverage of chemical space. To promote superior model development, the Division of Genetics and Mutagenesis, National Institute of Health Sciences, Japan (DGM/NIHS), conducted the Second Ames/QSAR International Challenge Project (2020-2022) as a successor to the First Project (2014-2017), with 21 teams from 11 countries participating. The DGM/NIHS provided a curated training dataset of approximately 12,000 chemicals and a trial dataset of approximately 1,600 chemicals, and each participating team predicted the Ames mutagenicity of each trial chemical using various Ames/QSAR models. The DGM/NIHS then provided the Ames test results for trial chemicals to assist in model improvement. Although overall model performance on the Second Project was not superior to that on the First, models from the eight teams participating in both projects achieved higher sensitivity than models from teams participating in only the Second Project. Thus, these evaluations have facilitated the development of QSAR models.

There is a pressing need to improve the efficiency of drug development, and nowhere is that need more clear than in the case of neglected diseases like malaria. The peculiarities of pyrimidine metabolism in Plasmodium species make inhibition of dihydroorotate dehydrogenase (DHODH) an attractive target for antimalarial drug design. By applying a pair of complementary quantitative structure–activity relationships derived for inhibition of a truncated, soluble form of the enzyme from Plasmodium falciparum (s-PfDHODH) to data from a large-scale phenotypic screen against cultured parasites, we were able to identify a class of antimalarial leads that inhibit the enzyme and abolish parasite growth in blood culture. Novel analogs extending that class were designed and synthesized with a goal of improving potency as well as the general pharmacokinetic and toxicological profiles. Their synthesis also represented an opportunity to prospectively validate our in silico property predictions. The seven analogs synthesized exhibited physicochemical properties in good agreement with prediction, and five of them were more active against P. falciparum growing in blood culture than any of the compounds in the published lead series. The particular analogs prepared did not inhibit s-PfDHODH in vitro, but advanced biological assays indicated that other examples from the class did inhibit intact PfDHODH bound to the mitochondrial membrane. The new analogs, however, killed the parasites by acting through some other, unidentified mechanism 24–48 h before PfDHODH inhibition would be expected to do so.

‘Deep learning’ is causing rapid technological changes in many fields of science, and conjectures about its potential for transforming everyone's work and lives is a matter of great debate. Unfortunately, it is all too easy to apply it as a ‘black box’ tool with little consideration of its potential limitations, especially when the data it is being applied to is less than perfect. In this Perspective, I try to put deep learning into a broader mechanistic and historical context by showing how it relates to older forms of artificial intelligence; by providing a general explanation of how it operates; and by exploring some of the challenges involved in its implementation. Examples wherein it has been applied to pest management problems are provided to illustrate how the technology works and the challenges deep learning faces. At least in the near term, its biggest impact on agrochemical development seems likely to come in automating the tedious work involved in assessing agrochemical efficacy, but getting there will require major investments in building large, well‐curated data sets to work from and in providing the expertise required to assess the resulting model predictions in real‐world scenarios. Deep learning may also come to complement the machine learning methodologies already available for use in pesticide discovery and development, but it seems unlikely to supplant them. © 2020 The Authors. Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

Deep Learning (DL) has gained wide adoption and different DL models have been deployed for an expanding number of applications. It is being used both for inference at the edge and for training in datacenters. Applications include image recognition, video analytics, pattern recognition in networking traffic, and many others. Different applications rely on different neural network models, and it has proven difficult to predict resource requirements for different models and applications. This leads to the nonsystematic and suboptimal selection of computational resources for DL applications resulting in overpaying for underutilized infrastructure or, even worse, the deployment of models on underpowered hardware and missed service level objectives. In this paper we present the DL Cookbook, a toolset that a) helps with benchmarking models on different hardware, b) guides the use of DL and c) provides reference designs. Automated benchmarking collects performance data for different DL workloads (training and inference with different models) on various hardware and software configurations. A web-based tool guides a choice of optimal hardware and software configuration via analysis of collected performance data and applying performance models. And finally, it offers reference hardware/software stacks for particular classes of deep learning workloads. This way the DL Cookbook helps both customers and hardware vendors match optimal DL models to the available hardware and vice versa, in case of acquisition, specify required hardware to models in question. Finally, DL Cookbook helps with reproducibility of results.

Currently, the submission guidelines for the Journal of Cheminformatics say it will “only publish research or software that is entirely reproducible by third parties.” They go on to specify that being reproducible means that anything essential to the conclusion of the paper be freely accessible and states that source code must be provided. I submit that this definition of reproducibility is too narrow—that a cheminformatics method can only truly be replicated by reimplementing it from a detailed, step-by-step high-level description to determine how reliably the algorithm per se does what it is intended to do.

Knowing the physicochemical and general biochemical properties of a compound is critical to understanding how it behaves in different biological environments and to anticipating what is likely to happen in situations where that behavior cannot be measured directly. Quantitative structure-property relationship (QSPR) models provide a way to predict those properties even before a compound has been synthesized simply by knowing what its structure would be. This chapter describes a general workflow for compiling the data upon which a useful QSPR model is built, curating it, evaluating that model’s performance, and then analyzing the predictive errors with an eye toward identifying systematic errors in the input data. The focus here is on models for the absorption, distribution, metabolism, and excretion (ADME) properties of drugs and toxins, but the considerations explored are general and applicable to any QSPR.

The back cover image, by Victor R. Aguilar, is based on the Mini‐Review by Robert D Clark, Predicting mammalian metabolism and toxicity of pesticides in silico , DOI: 10.1002/ps.4935 . image