PaccMannRL: De novo generation of hit-like anticancer molecules from transcriptomic data via reinforcement learning

Target-specific novel molecules with their recipe: Incorporating synthesizability in the design process

Application of Artificial intelligence (AI) in drug discovery has led to several success stories in recent times. While traditional methods mostly relied upon screening large chemical libraries for early-stage drug-design, de novo design can help identify novel target-specific molecules by sampling from a much larger chemical space. Although this has increased the possibility of finding diverse and novel molecules from previously unexplored chemical space, this has also posed a great challenge for medicinal chemists to synthesize at least some of the de novo designed novel molecules for experimental validation. To address this challenge, in this work, we propose a novel forward synthesis-based generative AI method, which is used to explore the synthesizable chemical space. The method uses a structure-based drug design framework, where the target protein structure and a target-specific seed fragment from co-crystal structures can be the initial inputs. A random fragment from a purchasable fragment library can also be the input if a target-specific fragment is unavailable. Then a template-based forward synthesis route prediction and molecule generation is performed in parallel using the Monte Carlo Tree Search (MCTS) method where, the subsequent fragments for molecule growth can again be obtained from a purchasable fragment library. The rewards for each iteration of MCTS are computed using a drug-target affinity (DTA) model based on the docking pose of the generated reaction intermediates at the binding site of the target protein of interest. With the help of the proposed method, it is now possible to overcome one of the major obstacles posed to the AI-based drug design approaches through the ability of the method to design novel target-specific synthesizable molecules.

TransGEM: a molecule generation model based on Transformer with gene expression data

MOTIVATION

It is difficult to generate new molecules with desirable bioactivity through ligand-based de novo drug design, and receptor-based de novo drug design is constrained by disease target information availability. The combination of artificial intelligence and phenotype-based de novo drug design can generate new bioactive molecules, independent from disease target information. Gene expression profiles can be used to characterize biological phenotypes. The Transformer model can be utilized to capture the associations between gene expression profiles and molecular structures due to its remarkable ability in processing contextual information.

RESULTS

We propose TransGEM (Transformer-based model from gene expression to molecules), which is a phenotype-based de novo drug design model. A specialized gene expression encoder is used to embed gene expression difference values between diseased cell lines and their corresponding normal tissue cells into TransGEM model. The results demonstrate that the TransGEM model can generate molecules with desirable evaluation metrics and property distributions. Case studies illustrate that TransGEM model can generate structurally novel molecules with good binding affinity to disease target proteins. The majority of genes with high attention scores obtained from TransGEM model are associated with the onset of the disease, indicating the potential of these genes as disease targets. Therefore, this study provides a new paradigm for de novo drug design, and it will promote phenotype-based drug discovery.

AVAILABILITY AND IMPLEMENTATION

The code is available at https://github.com/hzauzqy/TransGEM.

GLDM: hit molecule generation with constrained graph latent diffusion model

Discovering hit molecules with desired biological activity in a directed manner is a promising but profound task in computer-aided drug discovery. Inspired by recent generative AI approaches, particularly Diffusion Models (DM), we propose Graph Latent Diffusion Model (GLDM)-a latent DM that preserves both the effectiveness of autoencoders of compressing complex chemical data and the DM's capabilities of generating novel molecules. Specifically, we first develop an autoencoder to encode the molecular data into low-dimensional latent representations and then train the DM on the latent space to generate molecules inducing targeted biological activity defined by gene expression profiles. Manipulating DM in the latent space rather than the input space avoids complicated operations to map molecule decomposition and reconstruction to diffusion processes, and thus improves training efficiency. Experiments show that GLDM not only achieves outstanding performances on molecular generation benchmarks, but also generates samples with optimal chemical properties and potentials to induce desired biological activity.

Machine Learning Empowering Drug Discovery: Applications, Opportunities and Challenges

Drug discovery plays a critical role in advancing human health by developing new medications and treatments to combat diseases. How to accelerate the pace and reduce the costs of new drug discovery has long been a key concern for the pharmaceutical industry. Fortunately, by leveraging advanced algorithms, computational power and biological big data, artificial intelligence (AI) technology, especially machine learning (ML), holds the promise of making the hunt for new drugs more efficient. Recently, the Transformer-based models that have achieved revolutionary breakthroughs in natural language processing have sparked a new era of their applications in drug discovery. Herein, we introduce the latest applications of ML in drug discovery, highlight the potential of advanced Transformer-based ML models, and discuss the future prospects and challenges in the field.

Combined mechanistic modeling and machine-learning approaches in systems biology - A systematic literature review

BACKGROUND AND OBJECTIVE

Mechanistic-based Model simulations (MM) are an effective approach commonly employed, for research and learning purposes, to better investigate and understand the inherent behavior of biological systems. Recent advancements in modern technologies and the large availability of omics data allowed the application of Machine Learning (ML) techniques to different research fields, including systems biology. However, the availability of information regarding the analyzed biological context, sufficient experimental data, as well as the degree of computational complexity, represent some of the issues that both MMs and ML techniques could present individually. For this reason, recently, several studies suggest overcoming or significantly reducing these drawbacks by combining the above-mentioned two methods. In the wake of the growing interest in this hybrid analysis approach, with the present review, we want to systematically investigate the studies available in the scientific literature in which both MMs and ML have been combined to explain biological processes at genomics, proteomics, and metabolomics levels, or the behavior of entire cellular populations.

METHODS

Elsevier Scopus®, Clarivate Web of Science™ and National Library of Medicine PubMed® databases were enquired using the queries reported in Table 1, resulting in 350 scientific articles.

RESULTS

Only 14 of the 350 documents returned by the comprehensive search conducted on the three major online databases met our search criteria, i.e. present a hybrid approach consisting of the synergistic combination of MMs and ML to treat a particular aspect of systems biology.

CONCLUSIONS

Despite the recent interest in this methodology, from a careful analysis of the selected papers, it emerged how examples of integration between MMs and ML are already present in systems biology, highlighting the great potential of this hybrid approach to both at micro and macro biological scales.

De Novo Design of Molecules with Multiaction Potential from Differential Gene Expression using Variational Autoencoder

The modulating effect of chemical compounds and therapeutics on gene transcription is well-reported and has been intensively studied for both clinical and research purposes. Emerging research points toward the utility of drug-induced transcriptional alterations in de novo molecular design and highlights the idea of phenotype-matching an expression signature of interest to the structures being designed. In this work, we build an autoencoder-based generative model, BiCEV, around this concept. Our generative autoencoder has demonstrably generated a set of new molecules from gene expression input with notable validity (96%), uniqueness (98%), and internal diversity (0.77). Further, we attempted to validate BiCEV by testing the model on gene-knockdown profiles and combined signatures of synergistic drug pairs. From these investigations, we found the designed structures to be consistently high in collective quality. However, when their similarities to the supposed functional equivalents as determined by shared targets were considered, the findings were somewhat mixed. In spite of this, we believe the generative model merits further development in conjunction with in vitro corroboration to lend itself to being an assistive tool for drug discovery experts, particularly to support the initial stages of hit identification and lead optimization.

A Reinforcement Learning Framework to Discover Natural Flavor Molecules

Flavor is the focal point in the flavor industry, which follows social tendencies and behaviors. The research and development of new flavoring agents and molecules are essential in this field. However, the development of natural flavors plays a critical role in modern society. Considering this, the present work proposes a novel framework based on scientific machine learning to undertake an emerging problem in flavor engineering and industry. It proposes a combining system composed of generative and reinforcement learning models. Therefore, this work brings an innovative methodology to design new flavor molecules. The molecules were evaluated regarding synthetic accessibility, the number of atoms, and the likeness to a natural or pseudo-natural product. This work brings as contributions the implementation of a web scraper code to sample a flavors database and the integration of two scientific machine learning techniques in a complex system as a framework. The implementation of the complex system instead of the generative model by itself obtained 10% more molecules within the optimal results. The designed molecules obtained as an output of the reinforcement learning model’s generation were assessed regarding their existence or not in the market and whether they are already used in the flavor industry or not. Thus, we corroborated the potentiality of the framework presented for the search of molecules to be used in the development of flavor-based products.

Machine learning approaches to predict drug efficacy and toxicity in oncology

In recent years, there has been a surge of interest in using machine learning algorithms (MLAs) in oncology, particularly for biomedical applications such as drug discovery, drug repurposing, diagnostics, clinical trial design, and pharmaceutical production. MLAs have the potential to provide valuable insights and predictions in these areas by representing both the disease state and the therapeutic agents used to treat it. To fully utilize the capabilities of MLAs in oncology, it is important to understand the fundamental concepts underlying these algorithms and how they can be applied to assess the efficacy and toxicity of therapeutics. In this perspective, we lay out approaches to represent both the disease state and the therapeutic agents used by MLAs to derive novel insights and make relevant predictions.

Advances of Artificial Intelligence in Anti-Cancer Drug Design: A Review of the Past Decade

Anti-cancer drug design has been acknowledged as a complicated, expensive, time-consuming, and challenging task. How to reduce the research costs and speed up the development process of anti-cancer drug designs has become a challenging and urgent question for the pharmaceutical industry. Computer-aided drug design methods have played a major role in the development of cancer treatments for over three decades. Recently, artificial intelligence has emerged as a powerful and promising technology for faster, cheaper, and more effective anti-cancer drug designs. This study is a narrative review that reviews a wide range of applications of artificial intelligence-based methods in anti-cancer drug design. We further clarify the fundamental principles of these methods, along with their advantages and disadvantages. Furthermore, we collate a large number of databases, including the omics database, the epigenomics database, the chemical compound database, and drug databases. Other researchers can consider them and adapt them to their own requirements.

Potent antibiotic design via guided search from antibacterial activity evaluations

MOTIVATION

The emergence of drug-resistant bacteria makes the discovery of new antibiotics an urgent issue, but finding new molecules with the desired antibacterial activity is an extremely difficult task. To address this challenge, we established a framework, MDAGS (Molecular Design via Attribute-Guided Search), to optimize and generate potent antibiotic molecules.

RESULTS

By designing the antibacterial activity latent space and guiding the optimization of functional compounds based on this space, the model MDAGS can generate novel compounds with desirable antibacterial activity without the need for extensive expensive and time-consuming evaluations. Compared with existing antibiotics, candidate antibacterial compounds generated by MDAGS always possessed significantly better antibacterial activity and ensured high similarity. Furthermore, although without explicit constraints on similarity to known antibiotics, these candidate antibacterial compounds all exhibited the highest structural similarity to antibiotics of expected function in the DrugBank database query. Overall, our approach provides a viable solution to the problem of bacterial drug resistance.

AVAILABILITY AND IMPLEMENTATION

Code of the model and datasets can be downloaded from GitHub (https://github.com/LiangYu-Xidian/MDAGS).

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Leveraging molecular structure and bioactivity with chemical language models for de novo drug design

Generative chemical language models (CLMs) can be used for de novo molecular structure generation by learning from a textual representation of molecules. Here, we show that hybrid CLMs can additionally leverage the bioactivity information available for the training compounds. To computationally design ligands of phosphoinositide 3-kinase gamma (PI3Kγ), a collection of virtual molecules was created with a generative CLM. This virtual compound library was refined using a CLM-based classifier for bioactivity prediction. This second hybrid CLM was pretrained with patented molecular structures and fine-tuned with known PI3Kγ ligands. Several of the computer-generated molecular designs were commercially available, enabling fast prescreening and preliminary experimental validation. A new PI3Kγ ligand with sub-micromolar activity was identified, highlighting the method's scaffold-hopping potential. Chemical synthesis and biochemical testing of two of the top-ranked de novo designed molecules and their derivatives corroborated the model's ability to generate PI3Kγ ligands with medium to low nanomolar activity for hit-to-lead expansion. The most potent compounds led to pronounced inhibition of PI3K-dependent Akt phosphorylation in a medulloblastoma cell model, demonstrating efficacy of PI3Kγ ligands in PI3K/Akt pathway repression in human tumor cells. The results positively advocate hybrid CLMs for virtual compound screening and activity-focused molecular design.

Exploration of Chemical Space Guided by PixelCNN for Fragment-Based De Novo Drug Discovery

We report a novel framework for achieving fragment-based molecular design using pixel convolutional neural network (PixelCNN) combined with the simplified molecular input line entry system (SMILES) as molecular representation. While a widely used recurrent neural network (RNN) assumes monotonically decaying correlations in strings, PixelCNN captures a periodicity among characters of SMILES. Thus, PixelCNN provides us with a novel solution for the analysis of chemical space by extracting the periodicity of molecular structures that will be buried in SMILES. Moreover, this characteristic enables us to generate molecules by combining several simple building blocks, such as a benzene ring and side-chain structures, which contributes to the effective exploration of chemical space by step-by-step searching for molecules from a target fragment. In conclusion, PixelCNN could be a powerful approach focusing on the periodicity of molecules to explore chemical space for the fragment-based molecular design.

Predicting chemical structure using reinforcement learning with a stack-augmented conditional variational autoencoder

In this paper, a reinforcement learning model is proposed that can maximize the predicted binding affinity between a generated molecule and target proteins. The model used to generate molecules in the proposed model was the Stacked Conditional Variation AutoEncoder (Stack-CVAE), which acts as an agent in reinforcement learning so that the resulting chemical formulas have the desired chemical properties and show high binding affinity with specific target proteins. We generated 1000 chemical formulas using the chemical properties of sorafenib and the three target kinases of sorafenib. Then, we confirmed that Stack-CVAE generates more of the valid and unique chemical compounds that have the desired chemical properties and predicted binding affinity better than other generative models. More detailed analysis for 100 of the top scoring molecules show that they are novel ones not found in existing chemical databases. Moreover, they reveal significantly higher predicted binding affinity score for Raf kinases than for other kinases. Furthermore, they are highly druggable and synthesizable.

Gene-centric multi-omics integration with convolutional encoders for cancer drug response prediction

MOTIVATION

Tumor heterogeneity, including genetic and transcriptomic characteristics, can reduce the efficacy of anticancer pharmacological therapy, resulting in clinical variability in patient response to therapeutic medications. Multi-omics integration can allow in silico models to provide an additional perspective on a biological system.

METHODS

In this study, we propose a gene-centric multi-channel (GCMC) architecture to integrate multi-omics for predicting cancer drug response. GCMC transformed multi-omics profiles into a three-dimensional tensor with an additional dimension for omics types. GCMC's convolutional encoders captures multi-omics profiles for each gene and yields gene-centric features to predict drug responses.

RESULTS

We evaluated GCMC on various datasets, including The Cancer Genome Atlas (TCGA) patients, patient-derived xenografts (PDX) mice models, and the Genomics of Drug Sensitivity in Cancer (GDSC) cell line datasets. GCMC achieved better performance than baseline models, including single-omics models, in more than 75% of 265 drugs from GDSC cell line datasets. Furthermore, as for the clinical applicability of GCMC, it achieved the best performance on TCGA and PDX datasets in terms of both AUPR and AUC. We also analyzed models' capability of integrating multi-omics profiles by measuring the contribution ratio of omics types. GCMC can incorporate multi-omics profiles in various manners to enhance performance for each drug type. These results suggested that GCMC can improve performance and feature extraction capability by integrating multi-omics profiles in a gene-centric manner.

Computer especially AI-assisted drug virtual screening and design in traditional Chinese medicine

BACKGROUND

Traditional Chinese medicine (TCM), as a significant part of the global pharmaceutical science, the abundant molecular compounds it contains is a valuable potential source of designing and screening new drugs. However, due to the un-estimated quantity of the natural molecular compounds and diversity of the related problems drug discovery such as precise screening of molecular compounds or the evaluation of efficacy, physicochemical properties and pharmacokinetics, it is arduous for researchers to design or screen applicable compounds through old methods. With the rapid development of computer technology recently, especially artificial intelligence (AI), its innovation in the field of virtual screening contributes to an increasing efficiency and accuracy in the process of discovering new drugs.

PURPOSE

This study systematically reviewed the application of computational approaches and artificial intelligence in drug virtual filtering and devising of TCM and presented the potential perspective of computer-aided TCM development.

STUDY DESIGN

We made a systematic review following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Then screening the most typical articles for our research.

METHODS

The systematic review was performed by following the PRISMA guidelines. The databases PubMed, EMBASE, Web of Science, CNKI were used to search for publications that focused on computer-aided drug virtual screening and design in TCM.

RESULT

Totally, 42 corresponding articles were included in literature reviewing. Aforementioned studies were of great significance to the treatment and cost control of many challenging diseases such as COVID-19, diabetes, Alzheimer's Disease (AD), etc. Computational approaches and AI were widely used in virtual screening in the process of TCM advancing, which include structure-based virtual screening (SBVS) and ligand-based virtual screening (LBVS). Besides, computational technologies were also extensively applied in absorption, distribution, metabolism, excretion and toxicity (ADMET) prediction of candidate drugs and new drug design in crucial course of drug discovery.

CONCLUSIONS

The applications of computer and AI play an important role in the drug virtual screening and design in the field of TCM, with huge application prospects.

A comprehensive review on recent approaches for cancer drug discovery associated with artificial intelligence

Through the revolutionization of artificial intelligence (AI) technologies in clinical research, significant improvement is observed in diagnosis of cancer. Utilization of these AI technologies, such as machine and deep learning, is imperative for the discovery of novel anticancer drugs and improves existing/ongoing cancer therapeutics. However, building a model for complicated cancers and their types remains a challenge due to lack of effective therapeutics that hinder the establishment of effective computational tools. In this review, we exploit recent approaches and state-of-the-art in implementing AI methods for anticancer drug discovery, and discussed how advances in these applications need to be considered in the current cancer therapeutics. Considering the immense potential of AI, we explore molecular docking and their interactions to recognize metabolic activities that support drug design. Finally, we highlight corresponding strategies in applying machine and deep learning methods to various types of cancer with their pros and cons.

ChemistGA: A Chemical Synthesizable Accessible Molecular Generation Algorithm for Real-World Drug Discovery

Many deep learning (DL)-based molecular generative models have been proposed to design novel molecules. These models may perform well on benchmarks, but they usually do not take real-world constraints into account, such as available training data set, synthetic accessibility, and scaffold diversity in drug discovery. In this study, a new algorithm, ChemistGA, was proposed by combining the traditional heuristic algorithm with DL, in which the crossover of the traditional genetic algorithm (GA) was redefined by DL in conjunction with GA, and an innovative backcrossing operation was implemented to generate desired molecules. Our results clearly show that ChemistGA not only retains the strength of the traditional GA but also greatly enhances the synthetic accessibility and success rate of the generated molecules with desired properties. Calculations on the two benchmarks illustrate that ChemistGA achieves impressive performance among the state-of-the-art baselines, and it opens a new avenue for the application of generative models to real-world drug discovery scenarios.

De novo molecular design with deep molecular generative models for PPI inhibitors

We construct a protein-protein interaction (PPI) targeted drug-likeness dataset and propose a deep molecular generative framework to generate novel drug-likeness molecules from the features of the seed compounds. This framework gains inspiration from published molecular generative models, uses the key features associated with PPI inhibitors as input and develops deep molecular generative models for de novo molecular design of PPI inhibitors. For the first time, quantitative estimation index for compounds targeting PPI was applied to the evaluation of the molecular generation model for de novo design of PPI-targeted compounds. Our results estimated that the generated molecules had better PPI-targeted drug-likeness and drug-likeness. Additionally, our model also exhibits comparable performance to other several state-of-the-art molecule generation models. The generated molecules share chemical space with iPPI-DB inhibitors as demonstrated by chemical space analysis. The peptide characterization-oriented design of PPI inhibitors and the ligand-based design of PPI inhibitors are explored. Finally, we recommend that this framework will be an important step forward for the de novo design of PPI-targeted therapeutics.

Deep generative model for therapeutic targets using transcriptomic disease-associated data-USP7 case study

The generation of candidate hit molecules with the potential to be used in cancer treatment is a challenging task. In this context, computational methods based on deep learning have been employed to improve in silico drug design methodologies. Nonetheless, the applied strategies have focused solely on the chemical aspect of the generation of compounds, disregarding the likely biological consequences for the organism's dynamics. Herein, we propose a method to implement targeted molecular generation that employs biological information, namely, disease-associated gene expression data, to conduct the process of identifying interesting hits. When applied to the generation of USP7 putative inhibitors, the framework managed to generate promising compounds, with more than 90% of them containing drug-like properties and essential active groups for the interaction with the target. Hence, this work provides a novel and reliable method for generating new promising compounds focused on the biological context of the disease.

Machine learning approaches for biomolecular, biophysical, and biomaterials research

A fluent conversation with a virtual assistant, person-tailored news feeds, and deep-fake images created within seconds-all those things that have been unthinkable for a long time are now a part of our everyday lives. What these examples have in common is that they are realized by different means of machine learning (ML), a technology that has fundamentally changed many aspects of the modern world. The possibility to process enormous amount of data in multi-hierarchical, digital constructs has paved the way not only for creating intelligent systems but also for obtaining surprising new insight into many scientific problems. However, in the different areas of biosciences, which typically rely heavily on the collection of time-consuming experimental data, applying ML methods is a bit more challenging: Here, difficulties can arise from small datasets and the inherent, broad variability, and complexity associated with studying biological objects and phenomena. In this Review, we give an overview of commonly used ML algorithms (which are often referred to as "machines") and learning strategies as well as their applications in different bio-disciplines such as molecular biology, drug development, biophysics, and biomaterials science. We highlight how selected research questions from those fields were successfully translated into machine readable formats, discuss typical problems that can arise in this context, and provide an overview of how to resolve those encountered difficulties.

DeepTTA: a transformer-based model for predicting cancer drug response

Identifying new lead molecules to treat cancer requires more than a decade of dedicated effort. Before selected drug candidates are used in the clinic, their anti-cancer activity is generally validated by in vitro cellular experiments. Therefore, accurate prediction of cancer drug response is a critical and challenging task for anti-cancer drugs design and precision medicine. With the development of pharmacogenomics, the combination of efficient drug feature extraction methods and omics data has made it possible to use computational models to assist in drug response prediction. In this study, we propose DeepTTA, a novel end-to-end deep learning model that utilizes transformer for drug representation learning and a multilayer neural network for transcriptomic data prediction of the anti-cancer drug responses. Specifically, DeepTTA uses transcriptomic gene expression data and chemical substructures of drugs for drug response prediction. Compared to existing methods, DeepTTA achieved higher performance in terms of root mean square error, Pearson correlation coefficient and Spearman's rank correlation coefficient on multiple test sets. Moreover, we discovered that anti-cancer drugs bortezomib and dactinomycin provide a potential therapeutic option with multiple clinical indications. With its excellent performance, DeepTTA is expected to be an effective method in cancer drug design.

Generative machine learning for de novo drug discovery: A systematic review

Recent research on artificial intelligence indicates that machine learning algorithms can auto-generate novel drug-like molecules. Generative models have revolutionized de novo drug discovery, rendering the explorative process more efficient. Several model frameworks and input formats have been proposed to enhance the performance of intelligent algorithms in generative molecular design. In this systematic literature review of experimental articles and reviews over the last five years, machine learning models, challenges associated with computational molecule design along with proposed solutions, and molecular encoding methods are discussed. A query-based search of the PubMed, ScienceDirect, Springer, Wiley Online Library, arXiv, MDPI, bioRxiv, and IEEE Xplore databases yielded 87 studies. Twelve additional studies were identified via citation searching. Of the articles in which machine learning was implemented, six prominent algorithms were identified: long short-term memory recurrent neural networks (LSTM-RNNs), variational autoencoders (VAEs), generative adversarial networks (GANs), adversarial autoencoders (AAEs), evolutionary algorithms, and gated recurrent unit (GRU-RNNs). Furthermore, eight central challenges were designated: homogeneity of generated molecular libraries, deficient synthesizability, limited assay data, model interpretability, incapacity for multi-property optimization, incomparability, restricted molecule size, and uncertainty in model evaluation. Molecules were encoded either as strings, which were occasionally augmented using randomization, as 2D graphs, or as 3D graphs. Statistical analysis and visualization are performed to illustrate how approaches to machine learning in de novo drug design have evolved over the past five years. Finally, future opportunities and reservations are discussed.

Perplexity-Based Molecule Ranking and Bias Estimation of Chemical Language Models

Chemical language models (CLMs) can be employed to design molecules with desired properties. CLMs generate new chemical structures in the form of textual representations, such as the simplified molecular input line entry system (SMILES) strings. However, the quality of these de novo generated molecules is difficult to assess a priori. In this study, we apply the perplexity metric to determine the degree to which the molecules generated by a CLM match the desired design objectives. This model-intrinsic score allows identifying and ranking the most promising molecular designs based on the probabilities learned by the CLM. Using perplexity to compare "greedy" (beam search) with "explorative" (multinomial sampling) methods for SMILES generation, certain advantages of multinomial sampling become apparent. Additionally, perplexity scoring is performed to identify undesired model biases introduced during model training and allows the development of a new ranking system to remove those undesired biases.

Back translation for molecule generation

MOTIVATION

Molecule generation, which is to generate new molecules, is an important problem in bioinformatics. Typical tasks include generating molecules with given properties, molecular property improvement (i.e. improving specific properties of an input molecule), retrosynthesis (i.e. predicting the molecules that can be used to synthesize a target molecule), etc. Recently, deep-learning-based methods received more attention for molecule generation. The labeled data of bioinformatics is usually costly to obtain, but there are millions of unlabeled molecules. Inspired by the success of sequence generation in natural language processing with unlabeled data, we would like to explore an effective way of using unlabeled molecules for molecule generation.

RESULTS

We propose a new method, back translation for molecule generation, which is a simple yet effective semisupervised method. Let X be the source domain, which is the collection of properties, the molecules to be optimized, etc. Let Y be the target domain which is the collection of molecules. In particular, given a main task which is about to learn a mapping from the source domain X to the target domain Y, we first train a reversed model g for the Y to X mapping. After that, we use g to back translate the unlabeled data in Y to X and obtain more synthetic data. Finally, we combine the synthetic data with the labeled data and train a model for the main task. We conduct experiments on molecular property improvement and retrosynthesis, and we achieve state-of-the-art results on four molecule generation tasks and one retrosynthesis benchmark, USPTO-50k.

AVAILABILITY AND IMPLEMENTATION

Our code and data are available at https://github.com/fyabc/BT4MolGen.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Analysis of Training and Seed Bias in Small Molecules Generated with a Conditional Graph-Based Variational Autoencoder─Insights for Practical AI-Driven Molecule Generation

The application of deep learning to generative molecule design has shown early promise for accelerating lead series development. However, questions remain concerning how factors like training, data set, and seed bias impact the technology's utility to medicinal and computational chemists. In this work, we analyze the impact of seed and training bias on the output of an activity-conditioned graph-based variational autoencoder (VAE). Leveraging a massive, labeled data set corresponding to the dopamine D2 receptor, our graph-based generative model is shown to excel in producing desired conditioned activities and favorable unconditioned physical properties in generated molecules. We implement an activity-swapping method that allows for the activation, deactivation, or retention of activity of molecular seeds, and we apply independent deep learning classifiers to verify the generative results. Overall, we uncover relationships between noise, molecular seeds, and training set selection across a range of latent-space sampling procedures, providing important insights for practical AI-driven molecule generation.

MoleGuLAR: Molecule Generation Using Reinforcement Learning with Alternating Rewards

The design of new inhibitors for novel targets is a very important problem especially in the current scenario with the world being plagued by COVID-19. Conventional approaches such as high-throughput virtual screening require extensive combing through existing data sets in the hope of finding possible matches. In this study, we propose a computational strategy for de novo generation of molecules with high binding affinities to the specified target and other desirable properties for druglike molecules using reinforcement learning. A deep generative model built using a stack-augmented recurrent neural network initially trained to generate druglike molecules is optimized using reinforcement learning to start generating molecules with desirable properties like LogP, quantitative estimate of drug likeliness, topological polar surface area, and hydration free energy along with the binding affinity. For multiobjective optimization, we have devised a novel strategy in which the property being used to calculate the reward is changed periodically. In comparison to the conventional approach of taking a weighted sum of all rewards, this strategy shows an enhanced ability to generate a significantly higher number of molecules with desirable properties.

Active Site Sequence Representations of Human Kinases Outperform Full Sequence Representations for Affinity Prediction and Inhibitor Generation: 3D Effects in a 1D Model

Recent advances in deep learning have enabled the development of large-scale multimodal models for virtual screening and de novo molecular design. The human kinome with its abundant sequence and inhibitor data presents an attractive opportunity to develop proteochemometric models that exploit the size and internal diversity of this family of targets. Here, we challenge a standard practice in sequence-based affinity prediction models: instead of leveraging the full primary structure of proteins, each target is represented by a sequence of 29 discontiguous residues defining the ATP binding site. In kinase-ligand binding affinity prediction, our results show that the reduced active site sequence representation is not only computationally more efficient but consistently yields significantly higher performance than the full primary structure. This trend persists across different models, data sets, and performance metrics and holds true when predicting pIC₅₀ for both unseen ligands and kinases. Our interpretability analysis reveals a potential explanation for the superiority of the active site models: whereas only mild statistical effects about the extraction of three-dimensional (3D) interaction sites take place in the full sequence models, the active site models are equipped with an implicit but strong inductive bias about the 3D structure stemming from the discontiguity of the active sites. Moreover, in direct comparisons, our models perform similarly or better than previous state-of-the-art approaches in affinity prediction. We then investigate a de novo molecular design task and find that the active site provides benefits in the computational efficiency, but otherwise, both kinase representations yield similar optimized affinities (for both SMILES- and SELFIES-based molecular generators). Our work challenges the assumption that the full primary structure is indispensable for modeling human kinases.

De Novo Structure-Based Drug Design Using Deep Learning

In recent years, deep learning-based methods have emerged as promising tools for de novo drug design. Most of these methods are ligand-based, where an initial target-specific ligand data set is necessary to design potent molecules with optimized properties. Although there have been attempts to develop alternative ways to design target-specific ligand data sets, availability of such data sets remains a challenge while designing molecules against novel target proteins. In this work, we propose a deep learning-based method, where the knowledge of the active site structure of the target protein is sufficient to design new molecules. First, a graph attention model was used to learn the structure and features of the amino acids in the active site of proteins that are experimentally known to form protein-ligand complexes. Next, the learned active site features were used along with a pretrained generative model for conditional generation of new molecules. A bioactivity prediction model was then used in a reinforcement learning framework to optimize the conditional generative model. We validated our method against two well-studied proteins, Janus kinase 2 (JAK2) and dopamine receptor D2 (DRD2), where we produce molecules similar to the known inhibitors. The graph attention model could identify the probable key active site residues, which influenced the conditional molecule generator to design new molecules with pharmacophoric features similar to the known inhibitors.

Interpretable deep recommender system model for prediction of kinase inhibitor efficacy across cancer cell lines

Computational models for drug sensitivity prediction have the potential to significantly improve personalized cancer medicine. Drug sensitivity assays, combined with profiling of cancer cell lines and drugs become increasingly available for training such models. Multiple methods were proposed for predicting drug sensitivity from cancer cell line features, some in a multi-task fashion. So far, no such model leveraged drug inhibition profiles. Importantly, multi-task models require a tailored approach to model interpretability. In this work, we develop DEERS, a neural network recommender system for kinase inhibitor sensitivity prediction. The model utilizes molecular features of the cancer cell lines and kinase inhibition profiles of the drugs. DEERS incorporates two autoencoders to project cell line and drug features into 10-dimensional hidden representations and a feed-forward neural network to combine them into response prediction. We propose a novel interpretability approach, which in addition to the set of modeled features considers also the genes and processes outside of this set. Our approach outperforms simpler matrix factorization models, achieving R [Formula: see text] 0.82 correlation between true and predicted response for the unseen cell lines. The interpretability analysis identifies 67 biological processes that drive the cell line sensitivity to particular compounds. Detailed case studies are shown for PHA-793887, XMD14-99 and Dabrafenib.

TITAN: T-cell receptor specificity prediction with bimodal attention networks

MOTIVATION

The activity of the adaptive immune system is governed by T-cells and their specific T-cell receptors (TCR), which selectively recognize foreign antigens. Recent advances in experimental techniques have enabled sequencing of TCRs and their antigenic targets (epitopes), allowing to research the missing link between TCR sequence and epitope binding specificity. Scarcity of data and a large sequence space make this task challenging, and to date only models limited to a small set of epitopes have achieved good performance. Here, we establish a k-nearest-neighbor (K-NN) classifier as a strong baseline and then propose Tcr epITope bimodal Attention Networks (TITAN), a bimodal neural network that explicitly encodes both TCR sequences and epitopes to enable the independent study of generalization capabilities to unseen TCRs and/or epitopes.

RESULTS

By encoding epitopes at the atomic level with SMILES sequences, we leverage transfer learning and data augmentation to enrich the input data space and boost performance. TITAN achieves high performance in the prediction of specificity of unseen TCRs (ROC-AUC 0.87 in 10-fold CV) and surpasses the results of the current state-of-the-art (ImRex) by a large margin. Notably, our Levenshtein-based K-NN classifier also exhibits competitive performance on unseen TCRs. While the generalization to unseen epitopes remains challenging, we report two major breakthroughs. First, by dissecting the attention heatmaps, we demonstrate that the sparsity of available epitope data favors an implicit treatment of epitopes as classes. This may be a general problem that limits unseen epitope performance for sufficiently complex models. Second, we show that TITAN nevertheless exhibits significantly improved performance on unseen epitopes and is capable of focusing attention on chemically meaningful molecular structures.

AVAILABILITY AND IMPLEMENTATION

The code as well as the dataset used in this study is publicly available at https://github.com/PaccMann/TITAN.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

De novo molecular design and generative models

Molecular design strategies are integral to therapeutic progress in drug discovery. Computational approaches for de novo molecular design have been developed over the past three decades and, recently, thanks in part to advances in machine learning (ML) and artificial intelligence (AI), the drug discovery field has gained practical experience. Here, we review these learnings and present de novo approaches according to the coarseness of their molecular representation: that is, whether molecular design is modeled on an atom-based, fragment-based, or reaction-based paradigm. Furthermore, we emphasize the value of strong benchmarks, describe the main challenges to using these methods in practice, and provide a viewpoint on further opportunities for exploration and challenges to be tackled in the upcoming years.