Deep reinforcement learning for de novo drug design

Tracking protein kinase targeting advances: integrating QSAR into machine learning for kinase-targeted drug discovery

Protein kinases are vital drug targets, yet designing selective inhibitors is challenging, compounded by resistance and kinome complexity. This review explores Quantitative Structure-Activity Relationship (QSAR) modeling for kinase drug discovery, focusing on integrating traditional QSAR with machine learning (ML)-CNNs, RNNs-and structural data. Methods include structural databases, docking, and deep learning QSAR. Key findings show ML-integrated QSAR significantly improves selective inhibitor design for CDKs, JAKs, PIM kinases. The IDG-DREAM challenge exemplifies ML's potential for accurate kinase-inhibitor interaction prediction, outperforming traditional methods and enabling inhibitors with enhanced selectivity, efficacy, and resistance mitigation. QSAR combined with advanced computation and experimental data accelerates kinase drug discovery, offering transformative precision medicine potential. This review highlights deep learning-enhanced QSAR's novelty in automating feature extraction and capturing complex relationships, surpassing traditional QSAR, while emphasizing interpretability and experimental validation for clinical translation.

Artificial intelligence revolution in drug discovery: A paradigm shift in pharmaceutical innovation

Integrating artificial intelligence (AI) into drug discovery has revolutionized pharmaceutical innovation, addressing the challenges of traditional methods that are costly, time-consuming, and suffer from high failure rates. By utilizing machine learning (ML), deep learning (DL), and natural language processing (NLP), AI enhances various stages of drug development, including target identification, lead optimization, de novo drug design, and drug repurposing. AI tools, such as AlphaFold for protein structure prediction and AtomNet for structure-based drug design, have significantly accelerated the discovery process, improved efficiency and reduced costs. Success stories like Insilico Medicine's AI-designed molecule for idiopathic pulmonary fibrosis and BenevolentAI's identification of baricitinib for COVID-19 highlight AI's transformative potential. Additionally, AI enables the exploration of vast chemical spaces, optimization of clinical trials, and the identification of novel therapeutic targets, paving the way for precision medicine. However, challenges such as limited data accessibility, integration of diverse datasets, interpretability of AI models, and ethical concerns remain critical hurdles. Overcoming these limitations through enhanced algorithms, standardized databases, and interdisciplinary collaboration is essential. Overall, AI continues to reshape drug discovery, reducing timelines, increasing success rates, and driving the development of innovative and accessible therapies for unmet medical needs.

MGDM: Molecular generation using a multinomial diffusion model

Accurate analysis of molecular structures and the rapid generation of valid molecules remain significant challenges in De Novo drug design. In this study, we propose the Multinomial Generated Diffusion Model (MGDM) for molecular generation. This model leverages a multinomial diffusion framework to process discrete data, with a focus on learning the multinomial distribution inherent in the dataset. During the generation process, the model progressively denoises molecules, transitioning from a uniform noise distribution to ultimately produce valid molecular structures. Initially, we generate molecules unconditionally to expand the compound library. In the next phase, we focus on generating molecules with specific properties to assess the model's capacity for conditional generation. For this, we implement a classifier-free guidance strategy, which directs the diffusion model's task without the need for training separate classifier models. To validate the effectiveness of our framework, we conducted experiments using the Molecular Sets (MOSES) dataset. The results demonstrate that, compared to several state-of-the-art methods, MGDM generates valid molecules while achieving superior or comparable performance in terms of novelty and diversity.

Deep Reinforcement Learning-Based Self-Optimization of Flow Chemistry

The development of effective synthetic pathways is critical in many industrial sectors. The growing adoption of flow chemistry has opened new opportunities for more cost-effective and environmentally friendly manufacturing technologies. However, the development of effective flow chemistry processes is still hampered by labor- and experiment-intensive methodologies and poor or suboptimal performance. In this context, integrating advanced machine learning strategies into chemical process optimization can significantly reduce experimental burdens and enhance overall efficiency. This paper demonstrates the capabilities of deep reinforcement learning (DRL) as an effective self-optimization strategy for imine synthesis in flow, a key building block in many compounds such as pharmaceuticals and heterocyclic products. A deep deterministic policy gradient (DDPG) agent was designed to iteratively interact with the environment, the flow reactor, and learn how to deliver optimal operating conditions. A mathematical model of the reactor was developed based on new experimental data to train the agent and evaluate alternative self-optimization strategies. To optimize the DDPG agent's training performance, different hyperparameter tuning methods were investigated and compared, including trial-and-error and Bayesian optimization. Most importantly, a novel adaptive dynamic hyperparameter tuning was implemented to further enhance the training performance and optimization outcome of the agent. The performance of the proposed DRL strategy was compared against state-of-the-art gradient-free methods, namely SnobFit and Nelder-Mead. Finally, the outcomes of the different self-optimization strategies were tested experimentally. It was shown that the proposed DDPG agent has superior performance compared to its self-optimization counterparts. It offered better tracking of the global solution and reduced the number of required experiments by approximately 50 and 75% compared to Nelder-Mead and SnobFit, respectively. These findings hold significant promise for the chemical engineering community, offering a robust, efficient, and sustainable approach to optimizing flow chemistry processes and paving the way for broader integration of data-driven methods in process design and operation.

The search for superionic solid-state electrolytes using a physics-informed generative model

The discovery of superionic solid-state electrolytes for cation batteries is currently limited by the range of materials available in online materials databases. Generative artificial intelligence approaches have recently been applied to overcome this limitation and explore unknown stoichiometries and structures, but efficiently generating candidates that satisfy strict stability criteria remains challenging. Here we introduce a physics-informed hierarchical generative framework that leverages symmetry-aware crystallographic principles to systematically explore molecular configurations, lattice parameters, and bonding environments. Our approach integrates empirical physical constraints and reinforcement learning utilizing a hierarchical state representation to generate chemically valid and structurally stable candidates. We propose symmetry-aware hierarchical architecture for flow-based traversal with density (SHAFT-density) that ensures efficient exploration of the material search space, prioritizing low formation energy, molecular packing optimized for stability and conductivity, and enhanced electrochemical properties. We discovered new binary and ternary metastable phases, of which we find highly conductive LiBr, LiCl, Li2IBr, and Li3CBr2. These materials can either function as solid-state electrolyte materials or be part of solid-state electrolyte mixtures. Our results demonstrate the model's capability to identify stable, diverse, and potentially superionic compounds, offering promising candidates for developing next-generation solid-state electrolytes with improved characteristics.

Bridging BioSciences and technology: The impact of AI & GenAI in life sciences and agribusiness

The intersection of biosciences and technology has yielded transformative advancements, and Generative Artificial Intelligence (GenAI) started to stand at the forefront of this synergy. In the field of life sciences, GenAI is emerging as a catalyst, accelerating drug discovery by swiftly generating and predicting novel molecules. This expedites the identification of potential drug candidates, significantly reducing time and costs compared to traditional methods. Beyond drug discovery, GenAI contributes to protein folding predictions, genomics research, disease diagnosis and biomarker identification, enhancing our understanding of diseases and health conditions, fostering the development of personalized medicine. In agribusiness, GenAI proves instrumental in optimizing crop breeding and improving agricultural productivity. It can generate new crop varieties with desired traits by analyzing vast datasets comprising genomic and ecological information, addressing challenges such as disease resistance, improved yield, and enhanced nutritional content. Moreover, GenAI transcends traditional applications and extends its influence on synthetic biology, contributing to the design of novel enzymes and pathways. This opens avenues for bio-based manufacturing, renewable energy production, and environmental remediation. By harnessing the power of GenAI, the synergies between biosciences and technology accelerate innovation, improve efficiency, decrease costs, and address critical challenges. Conversely, the ethical considerations surrounding GenAI, especially Large Language Model (LLM) utilization in life sciences and agribusiness, such as data privacy, algorithmic bias, and the equitable distribution of benefits, must be addressed to ensure responsible and fair implementation, especially environment sustainability when utilizing this technology. This review article discusses the multifaceted impact of GenAI in a new era of advancements in life sciences and agribusiness.

Strategies and efforts in circumventing the emergence of antiviral resistance against conventional antivirals

Antiviral resistance stemming from rapid viral evolution and adaptation is a major challenge faced in treating viral infections. Here, we describe the mechanisms and factors underlying antiviral resistance and their implications to future drug development. Current improvements to conventional methods provide viable options to overcome antiviral resistance. Ongoing efforts in developing new antiviral strategies are also discussed. Examples from across virology are used to illustrate how virus evolution and antiviral therapy influence each other.

Evaluating AutoGrow4 - an open-source toolkit for semi-automated computer-aided drug discovery

INTRODUCTION

Drug discovery is a long and expensive process characterized by a high failure rate. To make this process more rational and efficient, scientists always look for new and better ways to design novel ligands for a target of interest. Among different approaches, de novo ones gained popularity in the last decade, thanks to their ability to efficiently explore the chemical space and their increasing reliability in generating high-quality compounds. Autogrow4 is open-source software for de novo drug design that generates ligands for a given target by exploiting a combination of a genetic algorithm and molecular docking calculations.

AREAS COVERED

In the present paper, the authors dissect this program's usefulness and limitations in generating new compounds from a pharmacodynamic and pharmacokinetic perspective. Specifically, this article examines all reported applications of the Autogrow code in the literature (as retrieved from the Scopus database) from the release of its first version in 2009 to the present.

EXPERT OPINION

In the hands of an expert molecular modeler, Autogrow4 is a useful tool for de novo ligand design. Its modular and open-source codebase offers many protocol customization features. The main downsides are limited control over the pharmacokinetic features of generated ligands and the bias toward high molecular weight compounds.

DiffMC-Gen: A Dual Denoising Diffusion Model for Multi-Conditional Molecular Generation

The precise and efficient design of potential drug molecules with diverse physicochemical properties has long been a critical challenge. In recent years, the emergence of various deep learning-based de novo molecular generation algorithms offered new directions to this issue, among which denoising diffusion models have demonstrated significant potential. However, previous methods often fail to simultaneously optimize multiple properties of candidate compounds, which may stem from directly employing nongeometric graph neural networks (GNNs), rendering them incapable of accurately capturing molecular topologic and geometric information. In this study, a dual denoising diffusion model is developed for multi-conditional molecular generation (DiffMC-Gen), which integrates both discrete and continuous features to enhance its ability to perceive 3D molecular structures. Additionally, it involves a multi-objective optimization strategy to simultaneously optimize multiple properties of the target molecule, including binding affinity, drug-likeness, synthesizability, and toxicity. From the perspectives of both 2D and 3D molecular generation, the molecules generated by DiffMC-Gen exhibit state-of-the-art (SOTA) performance in terms of novelty and uniqueness, meanwhile achieving comparable results to previous methods in drug-likeness and synthesizability. Furthermore, the generated molecules have well-predicted biological activity and druglike properties for three target proteins-LRRK2, HPK1, and GLP-1 receptor, while also maintaining high standards of validity, uniqueness, and novelty. These results underscore its potential for practical applications in drug design.

Moldrug algorithm for an automated ligand binding site exploration by 3D aware molecular enumerations

We present Moldrug, a computational tool for accelerating the hit-to-lead phase in structure-based drug design. Moldrug explores the chemical space using structural modifications suggested by the CReM library and by optimizing an adaptable fitness function with a genetic algorithm. Moldrug is complemented by Moldrug-Dashboard, a cross-platform and user-friendly graphical interface tailored for the analysis of Moldrug simulations. To illustrate Moldrug, we designed new potential inhibitors targeting the main protease (MPro) of SARS-CoV-2 by optimizing a consensus fitness function that balances binding affinity, drug-likeness, and synthetic accessibility. The designed molecules exhibited high chemical diversity. A subset of the designed molecules were ranked using MM/GBSA and alchemical binding free energy calculations, revealing predicted affinities as low as - 10 kcal mol - 1 . Moldrug is distributed as a Python package under the Apache 2.0 license. It offers pre-configured multi-parameter fitness functions for molecular design, while being highly adaptable for integrating functionalities from external software. Documentation and tutorials are available at https://moldrug.rtfd.io .

A Deep Generative Model for the Inverse Design of Transition Metal Ligands and Complexes

Deep generative models yielding transition metal complexes (TMCs) remain scarce despite the key role of these compounds in industrial catalytic processes, anticancer therapies, and the energy transition. Compared to drug discovery within the chemical space of organic molecules, TMCs pose further challenges, including the encoding of chemical bonds of higher complexity and the need to optimize multiple properties. In this work, we developed a generative model for the inverse design of transition metal ligands and complexes, based on the junction tree variational autoencoder (JT-VAE). After implementing a SMILES-based encoding of the metal-ligand bonds, the model was trained with the tmQMg-L ligand library, allowing for the generation of thousands of novel, highly diverse monodentate (κ1) and bidentate (κ2) ligands, including imines, phosphines, and carbenes. Further, the generated ligands were labeled with two target properties reflecting the stability and electron density of the associated homoleptic iridium TMCs: the HOMO-LUMO gap (ϵ) and the charge of the metal center (q Ir). This data was used to implement a conditional model that generated ligands from a prompt, with the single- or dual-objective of optimizing either or both the ϵ and q Ir properties and allowing for chemical interpretation based on the optimization trajectories. The optimizations also had an impact on other chemical properties, including ligand dissociation energies and oxidative addition barriers. A similar model was implemented to condition ligand generation by solubility and steric bulk.

Inverse design of metal-organic frameworks using deep dreaming approaches

Exploring the expansive and largely untapped chemical space of metal-organic frameworks (MOFs) holds promise for revolutionising the field of materials science. MOFs, hailed for their modular architecture, offer unmatched flexibility in customising functionalities to meet specific application needs. However, navigating this chemical space to identify optimal MOF structures poses a significant challenge. Traditional high-throughput computational screening (HTCS), while useful, is often limited by a distribution bias towards materials not aligned with the desired functionalities. To overcome these limitations, this study adopts a "deep dreaming" methodology to optimise MOFs in silico, aiming to generate structures with systematically shifted properties that are closer to target functionalities from the outset. Our approach integrates property prediction and structure optimisation within a single interpretable framework, leveraging a specialised chemical language model augmented with attention mechanisms. Focusing on a curated set of MOF properties critical to applications like carbon capture and energy storage, we demonstrate how deep dreaming can be utilised as a tool for targeted material design.

Token-Mol 1.0: tokenized drug design with large language models

The integration of large language models (LLMs) into drug design is gaining momentum; however, existing approaches often struggle to effectively incorporate three-dimensional molecular structures. Here, we present Token-Mol, a token-only 3D drug design model that encodes both 2D and 3D structural information, along with molecular properties, into discrete tokens. Built on a transformer decoder and trained with causal masking, Token-Mol introduces a Gaussian cross-entropy loss function tailored for regression tasks, enabling superior performance across multiple downstream applications. The model surpasses existing methods, improving molecular conformation generation by over 10% and 20% across two datasets, while outperforming token-only models by 30% in property prediction. In pocket-based molecular generation, it enhances drug-likeness and synthetic accessibility by approximately 11% and 14%, respectively. Notably, Token-Mol operates 35 times faster than expert diffusion models. In real-world validation, it improves success rates and, when combined with reinforcement learning, further optimizes affinity and drug-likeness, advancing AI-driven drug discovery.

Sculpting molecules in text-3D space: a flexible substructure aware framework for text-oriented molecular optimization

The integration of deep learning, particularly AI-Generated Content, with high-quality data derived from ab initio calculations has emerged as a promising avenue for transforming the landscape of scientific research. However, the challenge of designing molecular drugs or materials that incorporate multi-modality prior knowledge remains a critical and complex undertaking. Specifically, achieving a practical molecular design necessitates not only meeting the diversity requirements but also addressing structural and textural constraints with various symmetries outlined by domain experts. In this article, we present an innovative approach to tackle this inverse design problem by formulating it as a multi-modality guidance optimization task. Our proposed solution involves a textural-structure alignment symmetric diffusion framework for the implementation of molecular optimization tasks, namely 3DToMolo. 3DToMolo aims to harmonize diverse modalities including textual description features and graph structural features, aligning them seamlessly to produce molecular structures adhere to specified symmetric structural and textural constraints by experts in the field. Experimental trials across three guidance optimization settings have shown a superior hit optimization performance compared to state-of-the-art methodologies. Moreover, 3DToMolo demonstrates the capability to discover potential novel molecules, incorporating specified target substructures, without the need for prior knowledge. This work not only holds general significance for the advancement of deep learning methodologies but also paves the way for a transformative shift in molecular design strategies. 3DToMolo creates opportunities for a more nuanced and effective exploration of the vast chemical space, opening new frontiers in the development of molecular entities with tailored properties and functionalities.

Machine learning: Python tools for studying biomolecules and drug design

The increasing adoption of computational methods and artificial intelligence in scientific research has led to a growing interest in versatile tools like Python. In the fields of medical chemistry, biochemistry, and bioinformatics, Python has emerged as a key language for tackling complex challenges. It is used to solve various tasks, such as drug discovery, high-throughput and virtual screening, protein and genome analysis, and predicting drug efficacy. This review presents a list of tools for these tasks, including scripts, libraries, and ready-made programs, and serves as a starting point for scientists wishing to apply automation or optimization to routine tasks in medical chemistry and bioinformatics.

Comparing Molecules Generated by MMPDB and REINVENT4 with Ideas from Drug Discovery Design Teams

This study compares molecules designed by drug discovery project teams from the Sanders Tri-Institutional Therapeutics Discovery Institute with molecules generated by two computational tools: MMPDB and REINVENT4. Seven different test cases with diverse chemotypes are studied in order to explore the potential of these computational tools in complementing human expertise in the early stages of drug discovery. By comparing the molecular structures and properties generated by MMPDB and REINVENT4 to those designed by project design teams, we aim to assess the value of such tools. The results indicate that MMPDB and REINVENT4 cover regions of chemical space larger than those covered by ideas from the drug discovery project teams. However, the chemical spaces covered by the two methods are quite different, and neither method completely covers the chemical space identified by the drug discovery project teams. Thus, the computational methods are complementary to one another and to drug discovery project team ideation. Effective application of generative molecule design tools has the potential to accelerate the identification of novel therapeutic candidates by expanding the chemical space explored during drug discovery and enabling optimal exploration.

HMAMP: Designing Highly Potent Antimicrobial Peptides Using a Hypervolume-Driven Multiobjective Deep Generative Model

Antimicrobial peptides (AMPs) have exhibited unprecedented potential as biomaterials in combating multidrug-resistant bacteria, prompting the proposal of many excellent generative models. However, the multiobjective nature of AMP discovery is often overlooked, contributing to the high attrition rate of drug candidates. Here, we propose a novel approach termed hypervolume-driven multiobjective AMP design (HMAMP), which prioritizes the simultaneous optimization of multiattribute AMPs. By synergizing reinforcement learning and a gradient descent algorithm rooted in the hypervolume maximization concept, HMAMP effectively biases generative processes and mitigates the pattern collapse issue. Comparative experiments show that HMAMP significantly outperforms state-of-the-art methods in effectiveness and diversity. A knee-based decision strategy is then employed to fast screen candidates with favorable physicochemical properties, aligning with the enhanced antimicrobial activity and reduced side effects. Molecular visualization further elucidates structural and functional properties of the AMPs. Overall, HMAMP is an effective approach to traverse large and complex exploration spaces to search for idealism-realism trade-off AMPs.

ICVAE: Interpretable Conditional Variational Autoencoder for De Novo Molecular Design

Recent studies have demonstrated that machine learning-based generative models can create novel molecules with desirable properties. Among them, Conditional Variational Autoencoder (CVAE) is a powerful approach to generate molecules with desired physiochemical and pharmacological properties. However, the CVAE's latent space is still a black-box, making it difficult to understand the relationship between the latent space and molecular properties. To address this issue, we propose the Interpretable Conditional Variational Autoencoder (ICVAE), which introduces a modified loss function that correlates the latent value with molecular properties. ICVAE established a linear mapping between latent variables and molecular properties. This linearity is not only crucial for improving interpretability, by assigning clear semantic meaning to latent dimensions, but also provides a practical advantage. It enables direct manipulation of molecular attributes through simple coordinate shifts in latent space, rather than relying on opaque, black-box optimization algorithms. Our experimental results show that the ICVAE can linearly relate one or multiple molecular properties with the latent value and generate molecules with precise properties by controlling the latent values. The ICVAE's interpretability allows us to gain insight into the molecular generation process, making it a promising approach in drug discovery and material design.

New horizons at the interface of artificial intelligence and translational cancer research

Artificial intelligence (AI) is increasingly being utilized in cancer research as a computational strategy for analyzing multiomics datasets. Advances in single-cell and spatial profiling technologies have contributed significantly to our understanding of tumor biology, and AI methodologies are now being applied to accelerate translational efforts, including target discovery, biomarker identification, patient stratification, and therapeutic response prediction. Despite these advancements, the integration of AI into clinical workflows remains limited, presenting both challenges and opportunities. This review discusses AI applications in multiomics analysis and translational oncology, emphasizing their role in advancing biological discoveries and informing clinical decision-making. Key areas of focus include cellular heterogeneity, tumor microenvironment interactions, and AI-aided diagnostics. Challenges such as reproducibility, interpretability of AI models, and clinical integration are explored, with attention to strategies for addressing these hurdles. Together, these developments underscore the potential of AI and multiomics to enhance precision oncology and contribute to advancements in cancer care.

Advances in AI-based strategies and tools to facilitate natural product and drug development

Natural products and their derivatives have been important for treating diseases in humans, animals, and plants. However, discovering new structures from natural sources is still challenging. In recent years, artificial intelligence (AI) has greatly aided the discovery and development of natural products and drugs. AI facilitates to: connect genetic data to chemical structures or vice-versa, repurpose known natural products, predict metabolic pathways, and design and optimize metabolites biosynthesis. More recently, the emergence and improvement in neural networks such as deep learning and ensemble automated web based bioinformatics platforms have sped up the discovery process. Meanwhile, AI also improves the identification and structure elucidation of unknown compounds from raw data like mass spectrometry and nuclear magnetic resonance. This article reviews these AI-driven methods and tools, highlighting their practical applications and guide for efficient natural product discovery and drug development.

Exploring chemical space for "druglike" small molecules in the age of AI

The announcement of 2024 Nobel Prize in Chemistry to Alphafold has reiterated the role of AI in biology and mainly in the domain of "drug discovery". Till few years ago, structure-based drug design (SBDD) has been the preferred experimental design in many academic and pharmaceutical R and D divisions for developing novel therapeutics. However, with the advent of AI, the drug design field especially has seen a paradigm shift in its R&D across platforms. If "drug design" is a game, there are two main players, the small molecule drug and its target biomolecule, and the rules governing the game are mainly based on the interactions between these two players. In this brief review, we will be discussing our efforts in improving the state-of-the-art technology with respect to small molecules as well as in understanding the rules of the game. The review is broadly divided into five sections with the first section introducing the field and the challenges faced and the role of AI in this domain. In the second section, we describe some of the existing small molecule libraries developed in our labs and follow-up this section with a more recent knowledge-based resource available for public use. In section four, we describe some of the screening tools developed in our laboratories and are available for public use. Finally, section five delves into how domain knowledge is improving the utilization of AI in drug design. We provide three case studies from our work to illustrate this work. Finally, we conclude with our thoughts on the future scope of AI in drug design.

Integrating artificial intelligence in drug discovery and early drug development: a transformative approach

Artificial intelligence (AI) can transform drug discovery and early drug development by addressing inefficiencies in traditional methods, which often face high costs, long timelines, and low success rates. In this review we provide an overview of how to integrate AI to the current drug discovery and development process, as it can enhance activities like target identification, drug discovery, and early clinical development. Through multiomics data analysis and network-based approaches, AI can help to identify novel oncogenic vulnerabilities and key therapeutic targets. AI models, such as AlphaFold, predict protein structures with high accuracy, aiding druggability assessments and structure-based drug design. AI also facilitates virtual screening and de novo drug design, creating optimized molecular structures for specific biological properties. In early clinical development, AI supports patient recruitment by analyzing electronic health records and improves trial design through predictive modeling, protocol optimization, and adaptive strategies. Innovations like synthetic control arms and digital twins can reduce logistical and ethical challenges by simulating outcomes using real-world or virtual patient data. Despite these advancements, limitations remain. AI models may be biased if trained on unrepresentative datasets, and reliance on historical or synthetic data can lead to overfitting or lack generalizability. Ethical and regulatory issues, such as data privacy, also challenge the implementation of AI. In conclusion, in this review we provide a comprehensive overview about how to integrate AI into current processes. These efforts, although they will demand collaboration between professionals, and robust data quality, have a transformative potential to accelerate drug development.

Mol-AIR: Molecular Reinforcement Learning with Adaptive Intrinsic Rewards for Goal-Directed Molecular Generation

Optimizing techniques for discovering molecular structures with desired properties is crucial in artificial intelligence (AI)-based drug discovery. Combining deep generative models with reinforcement learning has emerged as an effective strategy for generating molecules with specific properties. Despite its potential, this approach is ineffective in exploring the vast chemical space and optimizing particular chemical properties. To overcome these limitations, we present Mol-AIR, a reinforcement learning-based framework using adaptive intrinsic rewards for effective goal-directed molecular generation. Mol-AIR leverages the strengths of both history-based and learning-based intrinsic rewards by exploiting random distillation network and counting-based strategies. In benchmark tests, Mol-AIR demonstrates improved performance over existing approaches in generating molecules having the desired properties, including penalized LogP, QED, and celecoxib similarity, without any prior knowledge. We believe that Mol-AIR represents a significant advancement in drug discovery, offering a more efficient path to discovering novel therapeutics.

CardioGenAI: a machine learning-based framework for re-engineering drugs for reduced hERG liability

The link between in vitro hERG ion channel inhibition and subsequent in vivo QT interval prolongation, a critical risk factor for the development of arrythmias such as Torsade de Pointes, is so well established that in vitro hERG activity alone is often sufficient to end the development of an otherwise promising drug candidate. It is therefore of tremendous interest to develop advanced methods for identifying hERG-active compounds in the early stages of drug development, as well as for proposing redesigned compounds with reduced hERG liability and preserved primary pharmacology. In this work, we present CardioGenAI, a machine learning-based framework for re-engineering both developmental and commercially available drugs for reduced hERG activity while preserving their pharmacological activity. The framework incorporates novel state-of-the-art discriminative models for predicting hERG channel activity, as well as activity against the voltage-gated NaV1.5 and CaV1.2 channels due to their potential implications in modulating the arrhythmogenic potential induced by hERG channel blockade. We applied the complete framework to pimozide, an FDA-approved antipsychotic agent that demonstrates high affinity to the hERG channel, and generated 100 refined candidates. Remarkably, among the candidates is fluspirilene, a compound which is of the same class of drugs as pimozide (diphenylmethanes) and therefore has similar pharmacological activity, yet exhibits over 700-fold weaker binding to hERG. Furthermore, we demonstrated the framework's ability to optimize hERG, NaV1.5 and CaV1.2 profiles of multiple FDA-approved compounds while maintaining the physicochemical nature of the original drugs. We envision that this method can effectively be applied to developmental compounds exhibiting hERG liabilities to provide a means of rescuing drug development programs that have stalled due to hERG-related safety concerns. Additionally, the discriminative models can also serve independently as effective components of virtual screening pipelines. We have made all of our software open-source at https://github.com/gregory-kyro/CardioGenAI to facilitate integration of the CardioGenAI framework for molecular hypothesis generation into drug discovery workflows.Scientific contributionThis work introduces CardioGenAI, an open-source machine learning-based framework designed to re-engineer drugs for reduced hERG liability while preserving their pharmacological activity. The complete CardioGenAI framework can be applied to developmental compounds exhibiting hERG liabilities to provide a means of rescuing drug discovery programs facing hERG-related challenges. In addition, the framework incorporates novel state-of-the-art discriminative models for predicting hERG, NaV1.5 and CaV1.2 channel activity, which can function independently as effective components of virtual screening pipelines.

Self-awareness of retrosynthesis via chemically inspired contrastive learning for reinforced molecule generation

The recent progress of deep generative models in modeling complex real-world data distributions has enabled the generation of novel compounds with potential therapeutic applications for various diseases. However, most studies fail to optimize the properties of generated molecules from the perspective of the intrinsic nature of chemical reactions. In this work, we propose a novel molecule generation model to overcome the limitation by deep reinforcement learning, in which an agent learns to optimize the properties of molecules initialized with a chemically inspired contrastive pretrained model. We finally assess the generation model by evaluating its ability to generate inhibitors against two prominent therapeutic targets in cancer treatment. Experimental results show that our model could generate 100% valid and novel structures and also exhibits superior performance in generating molecules with fewer structural alerts against several baselines. More importantly, the molecules generated by our proposed model show potent biological activities against ataxia telangiectasia and Rad3-related (ATR) and cyclin-dependent kinase 9 (CDK9) targets in wet-lab experiments.

Large-Scale Compartmental Model-Based Study of Preclinical Pharmacokinetic Data and Its Impact on Compound Triaging in Drug Discovery

Reliable and robust human dose prediction plays a pivotal role in drug discovery. The prediction of human dose requires proper modeling of preclinical intravenous (IV) pharmacokinetic (PK) data, which is usually achieved either through noncompartmental analysis (NCA) or compartmental analysis. While NCA is straightforward, it loses valuable information about the shape of the PK curves. In contrast, compartmental analysis offers a more comprehensive interpretation but poses challenges in scaling up for high-throughput applications in discovery. To address this challenge, we developed computational frameworks, termed compartmental PK (CPK) and automated dose prediction (ADP), to enable automated compartmental model-based IV PK data modeling, translation, and simulation for human dose prediction in compound triaging and optimization. With CPK and ADP, we analyzed compounds with data collected at the MRL between 2013 and 2023 to quantitatively characterize the impact of different PK modeling and simulation methods on human dose prediction. Our study revealed that despite minimal impact on estimating animal PK parameters, different methods significantly impacted predicted human dose, exposure, and Cmax, driven more by different simulation assumptions than by the PK modeling itself. CPK-ADP therefore enables us to efficiently perform complex human dose predictions on a large scale while integrating the latest and best information available on absorption, distribution, and clearance to support decision-making in discovery.

Reinforcement learning-based generative artificial intelligence for novel pesticide design

INTRODUCTION

Pesticides play a pivotal role in ensuring food security, and the development of green pesticides is an inevitable trend in global agricultural progress. Although deep learning-based generative models have revolutionized de novo drug design in pharmaceutical research, their application in pesticide research and development remains unexplored.

OBJECTIVES

This study aims to pioneer the application of generative artificial intelligence to pesticide design by proposing a reinforcement learning-based framework for obtaining pesticide-like molecules with high binding affinity.

METHODS

This framework comprises two key components: PestiGen-G, which systematically explores the pesticide-like chemical space using a character-based generative model coupled with the REINFORCE algorithm; and PestiGen-S, which combines a fragment-based generative model with the Monte Carlo Tree Search algorithm to generate molecules that stably bind to the specific target protein.

RESULTS

Experimental results show that the molecules generated by PestiGen have superior pesticide-likeness and binding affinity compared to those generated by existing methods. In addition, we employ an active learning strategy to reduce the false-positive rate of the generated molecules. Finally, through collaboration with domain experts, we successfully designed a novel 4-hydroxyphenylpyruvate dioxygenase inhibitor (YH23768) with favorable enzyme inhibition and herbicidal potency.

CONCLUSION

This proof-of-concept study highlights the utility of PestiGen as a valuable tool for pesticide design. The web server based on the model is freely available at https://dpai.ccnu.edu.cn/PestiGen/.

Meta-MolNet: A Cross-Domain Benchmark for Few Examples Drug Discovery

Predicting the pharmacological activity, toxicity, and pharmacokinetic properties of molecules is a central task in drug discovery. Existing machine learning methods are transferred from one resource rich molecular property to another data scarce property in the same scaffold dataset. However, existing models may produce fragile and highly uncertain predictions for new scaffold molecules. And these models were tested on different benchmarks, which seriously affected the quality of their evaluation results. In this article, we introduce Meta-MolNet, a collection of data benchmark and algorithms, which is a standard benchmark platform for measuring model generalization and uncertainty quantification capabilities. Meta-MolNet manages a wide range of molecular datasets with high ratio of molecules/scaffolds, which often leads to more difficult data shift and generalization problems. Furthermore, we propose a graph attention network based on cross-domain meta-learning, Meta-GAT, which uses bilevel optimization to learn meta-knowledge from the scaffold family molecular dataset in the source domain. Meta-GAT benefits from meta-knowledge that reduces the requirement of sample complexity to enable reliable predictions of new scaffold molecules in the target domain through internal iteration of a few examples. We evaluate existing methods as baselines for the community, and the Meta-MolNet benchmark demonstrates the effectiveness of measuring the proposed algorithm in domain generalization and uncertainty quantification. Extensive experiments demonstrate that the Meta-GAT model has state-of-the-art domain generalization performance and robustly estimates uncertainty under few examples constraints. By publishing AI-ready data, evaluation frameworks, and baseline results, we hope to see the Meta-MolNet suite become a comprehensive resource for the AI-assisted drug discovery community. Meta-MolNet is freely accessible at https://github.com/lol88/Meta-MolNet.

Advanced AI and ML frameworks for transforming drug discovery and optimization: With innovative insights in polypharmacology, drug repurposing, combination therapy and nanomedicine

Artificial Intelligence (AI) and Machine Learning (ML) are transforming drug discovery by overcoming traditional challenges like high costs, time-consuming, and frequent failures. AI-driven approaches streamline key phases, including target identification, lead optimization, de novo drug design, and drug repurposing. Frameworks such as deep neural networks (DNNs), convolutional neural networks (CNNs), and deep reinforcement learning (DRL) models have shown promise in identifying drug targets, optimizing delivery systems, and accelerating drug repurposing. Generative adversarial networks (GANs) and variational autoencoders (VAEs) aid de novo drug design by creating novel drug-like compounds with desired properties. Case studies, such as DDR1 kinase inhibitors designed using generative models and CDK20 inhibitors developed via structure-based methods, highlight AI's ability to produce highly specific therapeutics. Models like SNF-CVAE and DeepDR further advance drug repurposing by uncovering new therapeutic applications for existing drugs. Advanced ML algorithms enhance precision in predicting drug efficacy, toxicity, and ADME-Tox properties, reducing development costs and improving drug-target interactions. AI also supports polypharmacology by optimizing multi-target drug interactions and enhances combination therapy through predictions of drug synergies and antagonisms. In nanomedicine, AI models like CURATE.AI and the Hartung algorithm optimize personalized treatments by predicting toxicological risks and real-time dosing adjustments with high accuracy. Despite its potential, challenges like data quality, model interpretability, and ethical concerns must be addressed. High-quality datasets, transparent models, and unbiased algorithms are essential for reliable AI applications. As AI continues to evolve, it is poised to revolutionize drug discovery and personalized medicine, advancing therapeutic development and patient care.

Bayesian Flow Network Framework for Chemistry Tasks

In this work, we introduce ChemBFN, a language model that handles chemistry tasks based on Bayesian flow networks working with discrete data. A new accuracy schedule is proposed to improve sampling quality by significantly reducing reconstruction loss. We show evidence that our method is appropriate for generating molecules with satisfied diversity, even when a smaller number of sampling steps is used. A classifier-free guidance method is adapted for conditional generation. It is also worthwhile to point out that after generative training, our model can be fine-tuned on regression and classification tasks with state-of-the-art performance, which opens the gate of building all-in-one models in a single module style. Our model has been open sourced at https://github.com/Augus1999/bayesian-flow-network-for-chemistry.

Multiclass Synthetic Accessibility Prediction

Evaluating synthetic accessibility of in silico molecules is an integral component of the drug discovery process. While the application of machine learning models to predict whether small molecules are easy or hard to synthesize has gained attention recently, predetermined thresholds and data set imbalances present challenges for these binary classification approaches. In this study, we introduce a novel multiclass fold-ensembled classification approach to predict the minimum number of steps needed to synthesize a small molecule. By ensembling the base models trained on multiple stratified subsampled folds, this approach effectively mitigates the impact of class imbalance through probability aggregation or voting aggregation strategies. Additionally, we propose fuzzy evaluation metrics that account for practical tolerances in predictions, providing a more flexible and realistic assessment of model performance. Through experimentation on two reaction benchmark data sets, we demonstrate the effectiveness of our model in a multiclass synthetic accessibility prediction task and the superiority of our proposed method over six existing models in binary synthetic accessibility prediction tasks.

Dedenser: A Python Package for Clustering and Downsampling Chemical Libraries

The screening of chemical libraries is an essential starting point in the drug discovery process. While some researchers desire a more thorough screening of drug targets against a narrower scope of molecules, it is not uncommon for diverse screening sets to be favored during the early stages of drug discovery. However, a cost burden is associated with the screening of molecules, with potential drawbacks if particular areas of chemical space are needlessly overrepresented. To facilitate triaged sampling of chemical libraries and other collections of molecules, we have developed Dedenser, a tool for the downsampling of chemical clusters. Dedenser functions by reducing the membership of clusters within chemical point clouds while maintaining the initial topology or distribution in chemical space. Dedenser is a Python package that utilizes Hierarchical Density-Based Spatial Clustering of Applications with Noise to first identify clusters present in 3D chemical point clouds and then downsamples by applying Poisson disk sampling to clusters based on either their volume or density in chemical space. A command line interface tool and graphic user interface are available with Dedenser, which allow for the generation of chemical point clouds, using Mordred for QSAR descriptor calculations and uniform manifold approximation and projection for 3D embedding, as well as visualization. We hope that Dedenser will serve the community by enabling quick access to reduced collections of molecules that are representative of larger sets and selecting even distributions of molecules within clusters rather than single representative molecules from clusters. All code for Dedenser is open source and available at https://github.com/MSDLLCpapers/dedenser.

Simplicity within biological complexity

Motivation

Heterogeneous, interconnected, systems-level, molecular (multi-omic) data have become increasingly available and key in precision medicine. We need to utilize them to better stratify patients into risk groups, discover new biomarkers and targets, repurpose known and discover new drugs to personalize medical treatment. Existing methodologies are limited and a paradigm shift is needed to achieve quantitative and qualitative breakthroughs.

Results

In this perspective paper, we survey the literature and argue for the development of a comprehensive, general framework for embedding of multi-scale molecular network data that would enable their explainable exploitation in precision medicine in linear time. Network embedding methods (also called graph representation learning) map nodes to points in low-dimensional space, so that proximity in the learned space reflects the network's topology-function relationships. They have recently achieved unprecedented performance on hard problems of utilizing few omic data in various biomedical applications. However, research thus far has been limited to special variants of the problems and data, with the performance depending on the underlying topology-function network biology hypotheses, the biomedical applications, and evaluation metrics. The availability of multi-omic data, modern graph embedding paradigms and compute power call for a creation and training of efficient, explainable and controllable models, having no potentially dangerous, unexpected behaviour, that make a qualitative breakthrough. We propose to develop a general, comprehensive embedding framework for multi-omic network data, from models to efficient and scalable software implementation, and to apply it to biomedical informatics, focusing on precision medicine and personalized drug discovery. It will lead to a paradigm shift in the computational and biomedical understanding of data and diseases that will open up ways to solve some of the major bottlenecks in precision medicine and other domains.

Hybrid Unsupervised/Supervised Machine Learning for Identifying Molecular Structural Fingerprints From Ensemble Property

The ensemble properties of a system are obtained by averaging over the properties calculated for the various configurations it can have at a finite temperature and thus cannot be captured by a single molecular structure. Such ensemble properties are often important in material discovery. In designing new materials, the goal is to predict those ensemble structures that display a tailored property. However, mapping this average property to multiple structures introduces ambiguities and unreliable convergence in supervised machine learning. This presents a major obstacle in designing new materials. Here, we introduce a hybrid unsupervised/supervised learning method and demonstrate how to predict the structural parameters defining the conformers of a heterogeneous system, melanin, from its ensemble-averaged spectra. This also shows a new way to identify different structural fingerprints responsible for an ensemble-averaged superposition spectrum.

3DSMILES-GPT: 3D molecular pocket-based generation with token-only large language model

The generation of three-dimensional (3D) molecules based on target structures represents a cutting-edge challenge in drug discovery. Many existing approaches often produce molecules with invalid configurations, unphysical conformations, suboptimal drug-like qualities, limited synthesizability, and require extensive generation times. To address these challenges, we present 3DSMILES-GPT, a fully language-model-driven framework for 3D molecular generation that utilizes tokens exclusively. We treat both two-dimensional (2D) and 3D molecular representations as linguistic expressions, combining them through full-dimensional representations and pre-training the model on a vast dataset encompassing tens of millions of drug-like molecules. This token-only approach enables the model to comprehensively understand the 2D and 3D characteristics of large-scale molecules. Subsequently, we fine-tune the model using pair-wise structural data of protein pockets and molecules, followed by reinforcement learning to further optimize the biophysical and chemical properties of the generated molecules. Experimental results demonstrate that 3DSMILES-GPT generates molecules that comprehensively outperform existing methods in terms of binding affinity, drug-likeness (QED), and synthetic accessibility score (SAS). Notably, it achieves a 33% enhancement in the quantitative estimation of QED, meanwhile the binding affinity estimated by Vina docking maintaining its state-of-the-art performance. The generation speed is remarkably fast, with the average time approximately 0.45 seconds per generation, representing a threefold increase over the fastest existing methods. This innovative 3DSMILES-GPT approach has the potential to positively impact the generation of 3D molecules in drug discovery.

AI-based Virtual Screening of Traditional Chinese Medicine and the Discovery of Novel Inhibitors of TCTP

BACKGROUND

Translationally controlled tumour protein (TCTP) is associated with tumor diseases, such as breast cancer, and its inhibitor can reduce the growth of tumor cells. Unfortunately, there is currently no effective medication available for treating TCTP-related breast cancer.

OBJECTIVES

The objective of this study was to explore the inhibitor candidates among natural compounds for the treatment of breast cancer related to TCTP protein.

METHODS

To explore the potential inhibitors of TCTP, we first screened out four potential inhibitors in the Traditional Chinese Medicine (TCM) for cancer based on AI virtual screening using the docking method, and then revealed the interaction mechanism of TCTP and four candidate inhibitors from TCM with molecular docking and molecular dynamics (MD) methods.

RESULTS

Based on the conformational characteristics and the MD properties of the four leading compounds, we designed the new skeleton molecules with the AI method using MolAICal software. Our MD simulations have revealed that different small molecules bind to different sites of TCTP, but the flexible regions and the signaling pathways are almost the same, and the VDW and hydrophobic interactions are crucial in the interactions between TCTP and ligands.

CONCLUSION

We have proposed the candidate inhibitor of TCTP. Our study has provided a potential new method for exploring inhibitors from Traditional Chinese Medicine (TCM).

Researching public health datasets in the era of deep learning: a systematic literature review

Objective: Explore deep learning applications in predictive analytics for public health data, identify challenges and trends, and then understand the current landscape. Materials and Methods: A systematic literature review was conducted in June 2023 to search articles on public health data in the context of deep learning, published from the inception of medical and computer science databases through June 2023. The review focused on diverse datasets, abstracting applications, challenges, and advancements in deep learning. Results: 2004 articles were reviewed, identifying 14 disease categories. Observed trends include explainable-AI, patient embedding learning, and integrating different data sources and employing deep learning models in health informatics. Noted challenges were technical reproducibility and handling sensitive data. Discussion: There has been a notable surge in deep learning applications on public health data publications since 2015. Consistent deep learning applications and models continue to be applied across public health data. Despite the wide applications, a standard approach still does not exist for addressing the outstanding challenges and issues in this field. Conclusion: Guidelines are needed for applying deep learning and models in public health data to improve FAIRness, efficiency, transparency, comparability, and interoperability of research. Interdisciplinary collaboration among data scientists, public health experts, and policymakers is needed to harness the full potential of deep learning.

Role of Artificial Intelligence in Drug Discovery to Revolutionize the Pharmaceutical Industry: Resources, Methods and Applications

Traditional drug discovery methods such as wet-lab testing, validations, and synthetic techniques are time-consuming and expensive. Artificial Intelligence (AI) approaches have progressed to the point where they can have a significant impact on the drug discovery process. Using massive volumes of open data, artificial intelligence methods are revolutionizing the pharmaceutical industry. In the last few decades, many AI-based models have been developed and implemented in many areas of the drug development process. These models have been used as a supplement to conventional research to uncover superior pharmaceuticals expeditiously. AI's involvement in the pharmaceutical industry was used mostly for reverse engineering of existing patents and the invention of new synthesis pathways. Drug research and development to repurposing and productivity benefits in the pharmaceutical business through clinical trials. AI is studied in this article for its numerous potential uses. We have discussed how AI can be put to use in the pharmaceutical sector, specifically for predicting a drug's toxicity, bioactivity, and physicochemical characteristics, among other things. In this review article, we have discussed its application to a variety of problems, including de novo drug discovery, target structure prediction, interaction prediction, and binding affinity prediction. AI for predicting drug interactions and nanomedicines were also considered.

Revolutionising Neurological Therapeutics: Investigating Drug Repurposing Strategies

Repurposing drugs (DR) has become a viable approach to hasten the search for cures for neurodegenerative diseases (NDs). This review examines different off-target and on-target drug discovery techniques and how they might be used to find possible treatments for non-diagnostic depressions. Off-target strategies look at the known or unknown side effects of currently approved drugs for repositioning, whereas on-target strategies connect disease pathways to targets that can be treated with drugs. The review highlights the potential of experimental and computational methodologies, such as machine learning, proteomic techniques, network and genomics-based approaches, and in silico screening, in uncovering new drug-disease correlations. It also looks at difficulties and failed attempts at drug repurposing for NDs, highlighting the necessity of exact and standardised procedures to increase success rates. This review's objectives are to address the purpose of drug repurposing in human disorders, particularly neurological diseases, and to provide an overview of repurposing candidates that are presently undergoing clinical trials for neurological conditions, along with any possible causes and early findings. We then include a list of drug repurposing strategies, restrictions, and difficulties for upcoming research.

Status and Prospects of Research on Deep Learning-based De Novo Generation of Drug Molecules

Traditional molecular de novo generation methods, such as evolutionary algorithms, generate new molecules mainly by linking existing atomic building blocks. The challenging issues in these methods include difficulty in synthesis, failure to achieve desired properties, and structural optimization requirements. Advances in deep learning offer new ideas for rational and robust de novo drug design. Deep learning, a branch of machine learning, is more efficient than traditional methods for processing problems, such as speech, image, and translation. This study provides a comprehensive overview of the current state of research in de novo drug design based on deep learning and identifies key areas for further development. Deep learning-based de novo drug design is pivotal in four key dimensions. Molecular databases form the basis for model training, while effective molecular representations impact model performance. Common DL models (GANs, RNNs, VAEs, CNNs, DMs) generate drug molecules with desired properties. The evaluation metrics guide research directions by determining the quality and applicability of generated molecules. This abstract highlights the foundational aspects of DL-based de novo drug design, offering a concise overview of its multifaceted contributions. Consequently, deep learning in de novo molecule generation has attracted more attention from academics and industry. As a result, many deep learning-based de novo molecule generation types have been actively proposed.

Mining for Potent Inhibitors through Artificial Intelligence and Physics: A Unified Methodology for Ligand Based and Structure Based Drug Design

Determining the viability of a new drug molecule is a time- and resource-intensive task that makes computer-aided assessments a vital approach to rapid drug discovery. Here we develop a machine learning algorithm, iMiner, that generates novel inhibitor molecules for target proteins by combining deep reinforcement learning with real-time 3D molecular docking using AutoDock Vina, thereby simultaneously creating chemical novelty while constraining molecules for shape and molecular compatibility with target active sites. Moreover, through the use of various types of reward functions, we have introduced novelty in generative tasks for new molecules such as chemical similarity to a target ligand, molecules grown from known protein bound fragments, and creation of molecules that enforce interactions with target residues in the protein active site. The iMiner algorithm is embedded in a composite workflow that filters out Pan-assay interference compounds, Lipinski rule violations, uncommon structures in medicinal chemistry, and poor synthetic accessibility with options for cross-validation against other docking scoring functions and automation of a molecular dynamics simulation to measure pose stability. We also allow users to define a set of rules for the structures they would like to exclude during the training process and postfiltering steps. Because our approach relies only on the structure of the target protein, iMiner can be easily adapted for the future development of other inhibitors or small molecule therapeutics of any target protein.

Balancing the Functionality and Biocompatibility of Materials with a Deep-Learning-Based Inverse Design Framework

The rational design of molecules with the desired functionality presents a significant challenge in chemistry. Moreover, it is worth noting that making chemicals safe and sustainable is crucial to bringing them to the market. To address this, we propose a novel deep learning framework developed explicitly for inverse design of molecules with both functionality and biocompatibility. This innovative approach comprises two predictive models and one generative model, facilitating the targeted screening of novel molecules from created virtual chemical space. Our method's versatility is highlighted in the inverse design process, where it successfully generates molecules with specified motifs or composition, discovers synthetically accessible molecules, and jointly targets functional and safe properties beyond the training regime. The utility of this method is demonstrated in its ability to design ionic liquids (ILs) with enhanced antibacterial properties and reduced cytotoxicity, addressing the issue of balancing functionality and biocompatibility in molecular design.

AABBA Graph Kernel: Atom-Atom, Bond-Bond, and Bond-Atom Autocorrelations for Machine Learning

Graphs are one of the most natural and powerful representations available for molecules; natural because they have an intuitive correspondence to skeletal formulas, the language used by chemists worldwide, and powerful, because they are highly expressive both globally (molecular topology) and locally (atom and bond properties). Graph kernels are used to transform molecular graphs into fixed-length vectors, which, based on their capacity of measuring similarity, can be used as fingerprints for machine learning (ML). To date, graph kernels have mostly focused on the atomic nodes of the graph. In this work, we developed a graph kernel based on atom-atom, bond-bond, and bond-atom (AABBA) autocorrelations. The resulting vector representations were tested on regression ML tasks on a data set of transition metal complexes; a benchmark motivated by the higher complexity of these compounds relative to organic molecules. In particular, we tested different flavors of the AABBA kernel in the prediction of the energy barriers and bond distances of the Vaska's complex data set (Friederich et al., Chem. Sci., 2020, 11, 4584). For a variety of ML models, including neural networks, gradient boosting machines, and Gaussian processes, we showed that AABBA outperforms the baseline including only atom-atom autocorrelations. Dimensionality reduction studies also showed that the bond-bond and bond-atom autocorrelations yield many of the most relevant features. We believe that the AABBA graph kernel can accelerate the exploration of large chemical spaces and inspire novel molecular representations in which both atomic and bond properties play an important role.

Human-in-the-loop active learning for goal-oriented molecule generation

Machine learning (ML) systems have enabled the modelling of quantitative structure-property relationships (QSPR) and structure-activity relationships (QSAR) using existing experimental data to predict target properties for new molecules. These property predictors hold significant potential in accelerating drug discovery by guiding generative artificial intelligence (AI) agents to explore desired chemical spaces. However, they often struggle to generalize due to the limited scope of the training data. When optimized by generative agents, this limitation can result in the generation of molecules with artificially high predicted probabilities of satisfying target properties, which subsequently fail experimental validation. To address this challenge, we propose an adaptive approach that integrates active learning (AL) and iterative feedback to refine property predictors, thereby improving the outcomes of their optimization by generative AI agents. Our method leverages the Expected Predictive Information Gain (EPIG) criterion to select additional molecules for evaluation by an oracle. This process aims to provide the greatest reduction in predictive uncertainty, enabling more accurate model evaluations of subsequently generated molecules. Recognizing the impracticality of immediate wet-lab or physics-based experiments due to time and logistical constraints, we propose leveraging human experts for their cost-effectiveness and domain knowledge to effectively augment property predictors, bridging gaps in the limited training data. Empirical evaluations through both simulated and real human-in-the-loop experiments demonstrate that our approach refines property predictors to better align with oracle assessments. Additionally, we observe improved accuracy of predicted properties as well as improved drug-likeness among the top-ranking generated molecules. SCIENTIFIC CONTRIBUTION: We present an adaptable framework that integrates AL and human expertise to refine property predictors for goal-oriented molecule generation. This approach is robust to noise in human feedback and ensures that navigating chemical space with human-refined predictors leverages human insights to identify molecules that not only satisfy predicted property profiles but also score highly on oracle models. Additionally, it prioritizes practical characteristics such as drug-likeness, synthetic accessibility, and a favorable balance between exploring diverse chemical space and exploiting similarity to existing training data.

Navigating the oncology drug discovery and development process with programmes supported by the National Institutes of Health

The translation of basic drug discoveries from laboratories to clinical use presents substantial challenges. Factors such as insufficient funding, misdirected project focus, and inability to understand a drug's limitations or strengths contribute to the difficulty of this process. To address these issues, the National Institutes of Health (NIH) has established various resources dedicated to streamlining drug development. The NIH offers access to regularly curated databases encompassing categories like drug discovery, target discovery, genomics, proteomics, and clinical datasets. The NIH also provides access to key resources through various programmes, such as the Developmental Therapeutics Program, focusing on preclinical drug discovery and the Cancer Therapy Evaluation Program, which oversees clinical trial efforts for investigational agents. These resources might include funding opportunities, access to a network of scientific experts, and services to address gaps in scientific work. This Review explores the diverse platforms and resources available at the NIH and outlines how researchers can leverage them to expedite the drug development process.

Quantifying the use and potential benefits of artificial intelligence in scientific research

The rapid advancement of artificial intelligence (AI) is poised to reshape almost every line of work. Despite enormous efforts devoted to understanding AI's economic impacts, we lack a systematic understanding of the benefits to scientific research associated with the use of AI. Here we develop a measurement framework to estimate the direct use of AI and associated benefits in science. We find that the use and benefits of AI appear widespread throughout the sciences, growing especially rapidly since 2015. However, there is a substantial gap between AI education and its application in research, highlighting a misalignment between AI expertise supply and demand. Our analysis also reveals demographic disparities, with disciplines with higher proportions of women or Black scientists reaping fewer benefits from AI, potentially exacerbating existing inequalities in science. These findings have implications for the equity and sustainability of the research enterprise, especially as the integration of AI with science continues to deepen.

Advances in artificial intelligence-based technologies for increasing the quality of medical products

Artificial intelligence (AI) is a technology that combines machine learning (ML) and deep learning. It has numerous usages in the domains of medicine and other sciences. Artificial intelligence can forecast the behavior of a drug's target protein and predict its desired physicochemical qualities. AI's potential to enhance healthcare services offerings formerly unheard-of opportunities for cost reserves, enhanced overall clinical and patient outcomes. The recent development of research in the biomedical field, encompassing fields such as genomics, computational medicine, AI, and algorithms for learning, has led to the demand for novel technology, a fresh workforce, and new standards of practice set the stage for the revolution in healthcare. By connecting these health statistics with cutting-edge AI technologies, precise insights into patient treatment can be obtained. Moreover, AI can aid in the search for new drugs by foretelling the target protein's two-dimensional structure. In the current review, an overview of the latest AI-based technologies and how they may be employed to reduce product development time to market and snowballing product quality, cost-effectiveness, as well as security throughout the manufacturing process is detailed.

A systematic review of deep learning chemical language models in recent era

Discovering new chemical compounds with specific properties can provide advantages for fields that rely on materials for their development, although this task comes at a high cost in terms of complexity and resources. Since the beginning of the data age, deep learning techniques have revolutionized the process of designing molecules by analyzing and learning from representations of molecular data, greatly reducing the resources and time involved. Various deep learning approaches have been developed to date, using a variety of architectures and strategies, in order to explore the extensive and discontinuous chemical space, providing benefits for generating compounds with specific properties. In this study, we present a systematic review that offers a statistical description and comparison of the strategies utilized to generate molecules through deep learning techniques, utilizing the metrics proposed in Molecular Sets (MOSES) or Guacamol. The study included 48 articles retrieved from a query-based search of Scopus and Web of Science and 25 articles retrieved from citation search, yielding a total of 72 retrieved articles, of which 62 correspond to chemical language models approaches to molecule generation and other 10 retrieved articles correspond to molecular graph representations. Transformers, recurrent neural networks (RNNs), generative adversarial networks (GANs), Structured Space State Sequence (S4) models, and variational autoencoders (VAEs) are considered the main deep learning architectures used for molecule generation in the set of retrieved articles. In addition, transfer learning, reinforcement learning, and conditional learning are the most employed techniques for biased model generation and exploration of specific chemical space regions. Finally, this analysis focuses on the central themes of molecular representation, databases, training dataset size, validity-novelty trade-off, and performance of unbiased and biased chemical language models. These themes were selected to conduct a statistical analysis utilizing graphical representation and statistical tests. The resulting analysis reveals the main challenges, advantages, and opportunities in the field of chemical language models over the past four years.

Transferable and Interpretable Prediction of Site-Specific Dehydrogenation Reaction Rate Constants with NMR Spectra

Accurate and efficient determination of site-specific reaction rate constants over a wide temperature range remains challenging, both experimentally and theoretically. Taking the dehydrogenation reaction as an example, our study addresses this issue by an innovative combination of machine learning techniques and cost-effective NMR spectra. Through descriptor screening, we identified a minimal set of NMR chemical shifts that can effectively determine reaction rate constants. The constructed model performs exceptionally well on theoretical data sets and demonstrates impressive generalization capabilities, extending from small molecules to larger ones. Furthermore, this model shows outstanding performance when applied to limited experimental data sets, highlighting its robust applicability and transferability. Utilizing the Sure Independence Screening and Sparsifying Operator (SISSO) algorithm, we also present an interpretable rate constant-temperature-NMR (k-T-NMR) relationship with a mathematical formula. This study reveals the great potential of combining machine learning with easily accessible spectroscopic descriptors in the study of reaction kinetics, enabling the rapid determination of reaction rate constants and promoting our understanding of reactivity.

De Novo Drug Design by Multi-Objective Path Consistency Learning With Beam A* Search

Generating high-quality and drug-like molecules from scratch within the expansive chemical space presents a significant challenge in the field of drug discovery. In prior research, value-based reinforcement learning algorithms have been employed to generate molecules with multiple desired properties iteratively. The immediate reward was defined as the evaluation of intermediate-state molecules at each step, and the learning objective would be maximizing the expected cumulative evaluation scores for all molecules along the generative path. However, this definition of the reward was misleading, as in reality, the optimization target should be the evaluation score of only the final generated molecule. Furthermore, in previous works, randomness was introduced into the decision-making process, enabling the generation of diverse molecules but no longer pursuing the maximum future rewards. In this paper, immediate reward is defined as the improvement achieved through the modification of the molecule to maximize the evaluation score of the final generated molecule exclusively. Originating from the A search, path consistency (PC), i.e., values on one optimal path should be identical, is employed as the objective function in the update of the value estimator to train a multi-objective de novo drug designer. By incorporating the value into the decision-making process of beam search, the DrugBA algorithm is proposed to enable the large-scale generation of molecules that exhibit both high quality and diversity. Experimental results demonstrate a substantial enhancement over the state-of-the-art algorithm QADD in multiple molecular properties of the generated molecules.