ChemSpaceAL: An Efficient Active Learning Methodology Applied to Protein-Specific Molecular Generation

The incredible capabilities of generative artificial intelligence models have inevitably led to their application in the domain of drug discovery. Within this domain, the vastness of chemical space motivates the development of more efficient methods for identifying regions with molecules that exhibit desired characteristics. In this work, we present a computationally efficient active learning methodology and demonstrate its applicability to targeted molecular generation. When applied to c-Abl kinase, a protein with FDA-approved small-molecule inhibitors, the model learns to generate molecules similar to the inhibitors without prior knowledge of their existence and even reproduces two of them exactly. We also show that the methodology is effective for a protein without any commercially available small-molecule inhibitors, the HNH domain of the CRISPR-associated protein 9 (Cas9) enzyme. To facilitate implementation and reproducibility, we made all of our software available through the open-source ChemSpaceAL Python package.

On the difficulty of validating molecular generative models realistically: a case study on public and proprietary data

While a multitude of deep generative models have recently emerged there exists no best practice for their practically relevant validation. On the one hand, novel de novo-generated molecules cannot be refuted by retrospective validation (so that this type of validation is biased); but on the other hand prospective validation is expensive and then often biased by the human selection process. In this case study, we frame retrospective validation as the ability to mimic human drug design, by answering the following question: Can a generative model trained on early-stage project compounds generate middle/late-stage compounds de novo? To this end, we used experimental data that contains the elapsed time of a synthetic expansion following hit identification from five public (where the time series was pre-processed to better reflect realistic synthetic expansions) and six in-house project datasets, and used REINVENT as a widely adopted RNN-based generative model. After splitting the dataset and training REINVENT on early-stage compounds, we found that rediscovery of middle/late-stage compounds was much higher in public projects (at 1.60%, 0.64%, and 0.21% of the top 100, 500, and 5000 scored generated compounds) than in in-house projects (where the values were 0.00%, 0.03%, and 0.04%, respectively). Similarly, average single nearest neighbour similarity between early- and middle/late-stage compounds in public projects was higher between active compounds than inactive compounds; however, for in-house projects the converse was true, which makes rediscovery (if so desired) more difficult. We hence show that the generative model recovers very few middle/late-stage compounds from real-world drug discovery projects, highlighting the fundamental difference between purely algorithmic design and drug discovery as a real-world process. Evaluating de novo compound design approaches appears, based on the current study, difficult or even impossible to do retrospectively.Scientific Contribution This contribution hence illustrates aspects of evaluating the performance of generative models in a real-world setting which have not been extensively described previously and which hopefully contribute to their further future development.

Deep learning driven de novo drug design based on gastric proton pump structures

Existing drugs often suffer in their effectiveness due to detrimental side effects, low binding affinity or pharmacokinetic problems. This may be overcome by the development of distinct compounds. Here, we exploit the rich structural basis of drug-bound gastric proton pump to develop compounds with strong inhibitory potency, employing a combinatorial approach utilizing deep generative models for de novo drug design with organic synthesis and cryo-EM structural analysis. Candidate compounds that satisfy pharmacophores defined in the drug-bound proton pump structures, were designed in silico utilizing our deep generative models, a workflow termed Deep Quartet. Several candidates were synthesized and screened according to their inhibition potencies in vitro, and their binding poses were in turn identified by cryo-EM. Structures reaching up to 2.10 Å resolution allowed us to evaluate and re-design compound structures, heralding the most potent compound in this study, DQ-18 (N-methyl-4-((2-(benzyloxy)-5-chlorobenzyl)oxy)benzylamine), which shows a Ki value of 47.6 nM. Further high-resolution cryo-EM analysis at 2.08 Å resolution unambiguously determined the DQ-18 binding pose. Our integrated approach offers a framework for structure-based de novo drug development based on the desired pharmacophores within the protein structure.

Machine learning & deep learning in data-driven decision making of drug discovery and challenges in high-quality data acquisition in the pharmaceutical industry

Predicting novel small molecule bioactivities for the target deconvolution, hit-to-lead optimization in drug discovery research, requires molecular representation. Previous reports have demonstrated that machine learning (ML) and deep learning (DL) have substantial implications in virtual screening, peptide synthesis, drug ADMET screening and biomarker discovery. These strategies can increase the positive outcomes in the drug discovery process without false-positive rates and can be achieved in a cost-effective way with a minimum duration of time by high-quality data acquisition. This review substantially discusses the recent updates in AI tools as cheminformatics application in medicinal chemistry for the data-driven decision making of drug discovery and challenges in high-quality data acquisition in the pharmaceutical industry while improving small-molecule bioactivities and properties.

Inhibitors of Discoidin Domain Receptor (DDR) Kinases for Cancer and Inflammation

The discoidin domain receptor tyrosine kinases DDR1 and DDR2 are distinguished from other kinase enzymes by their extracellular domains, which interact with collagen rather than with peptidic growth factors, before initiating signaling via tyrosine phosphorylation. They share significant sequence and structural homology with both the c-Kit and Bcr-Abl kinases, and so many inhibitors of those kinases are also effective. Nevertheless, there has been an extensive research effort to develop potent and specific DDR inhibitors. A key interaction for many of these compounds is H-bonding to Met-704 in a hydrophobic pocket of the DDR enzyme. The most widespread use of DDR inhibitors has been for cancer therapy, but they have also shown effectiveness in animal models of inflammatory conditions such as Alzheimer's and Parkinson's diseases, and in chronic renal failure and glomerulonephritis.

V-Dock: Fast Generation of Novel Drug-like Molecules Using Machine-Learning-Based Docking Score and Molecular Optimization

We propose a computational workflow to design novel drug-like molecules by combining the global optimization of molecular properties and protein-ligand docking with machine learning. However, most existing methods depend heavily on experimental data, and many targets do not have sufficient data to train reliable activity prediction models. To overcome this limitation, protein-ligand docking calculations must be performed using the limited data available. Such docking calculations during molecular generation require considerable computational time, preventing extensive exploration of the chemical space. To address this problem, we trained a machine-learning-based model that predicted the docking energy using SMILES to accelerate the molecular generation process. Docking scores could be accurately predicted using only a SMILES string. We combined this docking score prediction model with the global molecular property optimization approach, MolFinder, to find novel molecules exhibiting the desired properties with high values of predicted docking scores. We named this design approach V-dock. Using V-dock, we efficiently generated many novel molecules with high docking scores for a target protein, a similarity to the reference molecule, and desirable drug-like and bespoke properties, such as QED. The predicted docking scores of the generated molecules were verified by correlating them with the actual docking scores.

Current trends in computer aided drug design and a highlight of drugs discovered via computational techniques: A review

Computer-aided drug design (CADD) is one of the pivotal approaches to contemporary pre-clinical drug discovery, and various computational techniques and software programs are typically used in combination, in a bid to achieve the desired outcome. Several approved drugs have been developed with the aid of CADD. On SciFinder®, we evaluated more than 600 publications through systematic searching and refining, using the terms, virtual screening; software methods; computational studies and publication year, in order to obtain data concerning particular aspects of CADD. The primary focus of this review was on the databases screened, virtual screening and/or molecular docking software program used. Furthermore, we evaluated the studies that subsequently performed molecular dynamics (MD) simulations and we reviewed the software programs applied, the application of density functional theory (DFT) calculations and experimental assays. To represent the latest trends, the most recent data obtained was between 2015 and 2020, consequently the most frequently employed techniques and software programs were recorded. Among these, the ZINC database was the most widely preferred with an average use of 31.2%. Structure-based virtual screening (SBVS) was the most prominently used type of virtual screening and it accounted for an average of 57.6%, with AutoDock being the preferred virtual screening/molecular docking program with 41.8% usage. Following the screening process, 38.5% of the studies performed MD simulations to complement the virtual screening and GROMACS with 39.3% usage, was the popular MD software program. Among the computational techniques, DFT was the least applied whereby it only accounts for 0.02% average use. An average of 36.5% of the studies included reports on experimental evaluations following virtual screening. Ultimately, since the inception and application of CADD in pre-clinical drug discovery, more than 70 approved drugs have been discovered, and this number is steadily increasing over time.

Reinforcement learning to boost molecular docking upon protein conformational ensemble

Intrinsically disordered proteins (IDPs) are widely involved in human diseases and thus are attractive therapeutic targets. In practice, however, it is computationally prohibitive to dock large ligand libraries to thousands and tens of thousands of conformations. Here, we propose a reversible upper confidence bound (UCB) algorithm for the virtual screening of IDPs to address the influence of the conformation ensemble. The docking process is dynamically arranged so that attempts are focused near the boundary to separate top ligands from the bulk accurately. It is demonstrated in the example of transcription factor c-Myc that the average docking number per ligand can be greatly reduced while the performance is merely slightly affected. This study suggests that reinforcement learning is highly efficient in solving the bottleneck of virtual screening due to the conformation ensemble in the rational drug design of IDPs.

Design and Synthesis of DDR1 Inhibitors with a Desired Pharmacophore Using Deep Generative Models

Discoidin domain receptor 1 (DDR1) inhibitors with a desired pharmacophore were designed using deep generative models (DGMs). DDR1 is a receptor tyrosine kinase activated by matrix collagens and implicated in diseases such as cancer, fibrosis and hypoxia. Herein we describe the synthesis and inhibitory activity of compounds generated from DGMs. Three compounds were found to have sub-micromolar inhibitory activity. The most potent of which, compound 3 (N-(4-chloro-3-((pyridin-3-yloxy)methyl)phenyl)-3-(trifluoromethyl)benzamide), had an IC50 value of 92.5 nM. Furthermore, these compounds were predicted to interact with DDR1, which have a desired pharmacophore derived from a known DDR1 inhibitor. The results of synthesis and experiments indicated that our de novo design strategy is practical for hit identification and scaffold hopping.

Deep learning and generative methods in cheminformatics and chemical biology: navigating small molecule space intelligently

The number of 'small' molecules that may be of interest to chemical biologists - chemical space - is enormous, but the fraction that have ever been made is tiny. Most strategies are discriminative, i.e. have involved 'forward' problems (have molecule, establish properties). However, we normally wish to solve the much harder generative or inverse problem (describe desired properties, find molecule). 'Deep' (machine) learning based on large-scale neural networks underpins technologies such as computer vision, natural language processing, driverless cars, and world-leading performance in games such as Go; it can also be applied to the solution of inverse problems in chemical biology. In particular, recent developments in deep learning admit the in silico generation of candidate molecular structures and the prediction of their properties, thereby allowing one to navigate (bio)chemical space intelligently. These methods are revolutionary but require an understanding of both (bio)chemistry and computer science to be exploited to best advantage. We give a high-level (non-mathematical) background to the deep learning revolution, and set out the crucial issue for chemical biology and informatics as a two-way mapping from the discrete nature of individual molecules to the continuous but high-dimensional latent representation that may best reflect chemical space. A variety of architectures can do this; we focus on a particular type known as variational autoencoders. We then provide some examples of recent successes of these kinds of approach, and a look towards the future.

DeepGraphMolGen, a multi-objective, computational strategy for generating molecules with desirable properties: a graph convolution and reinforcement learning approach

We address the problem of generating novel molecules with desired interaction properties as a multi-objective optimization problem. Interaction binding models are learned from binding data using graph convolution networks (GCNs). Since the experimentally obtained property scores are recognised as having potentially gross errors, we adopted a robust loss for the model. Combinations of these terms, including drug likeness and synthetic accessibility, are then optimized using reinforcement learning based on a graph convolution policy approach. Some of the molecules generated, while legitimate chemically, can have excellent drug-likeness scores but appear unusual. We provide an example based on the binding potency of small molecules to dopamine transporters. We extend our method successfully to use a multi-objective reward function, in this case for generating novel molecules that bind with dopamine transporters but not with those for norepinephrine. Our method should be generally applicable to the generation in silico of molecules with desirable properties.