Early Stages Of Drug Discovery Research Articles

Accelerated hit identification with target evaluation, deep learning and automated labs: prospective validation in IRAK1.

Target identification and hit identification can be transformed through the application of biomedical knowledge analysis, AI-driven virtual screening and robotic cloud lab systems. However there are few prospective studies that evaluate the efficacy of such integrated approaches. We synergistically integrate our in-house-developed target evaluation (SpectraView) and deep-learning-driven virtual screening (HydraScreen) tools with an automated robotic cloud lab designed explicitly for ultra-high-throughput screening, enabling us to validate these platforms experimentally. By employing our target evaluation tool to select IRAK1 as the focal point of our investigation, we prospectively validate our structure-based deep learning model. We can identify 23.8% of all IRAK1 hits within the top 1% of ranked compounds. The model outperforms traditional virtual screening techniques and offers advanced features such as ligand pose confidence scoring. Simultaneously, we identify three potent (nanomolar) scaffolds from our compound library, 2 of which represent novel candidates for IRAK1 and hold promise for future development. This study provides compelling evidence for SpectraView and HydraScreen to provide a significant acceleration in the processes of target identification and hit discovery. By leveraging Ro5's HydraScreen and Strateos' automated labs in hit identification for IRAK1, we show how AI-driven virtual screening with HydraScreen could offer high hit discovery rates and reduce experimental costs. We present an innovative platform that leverages Knowledge graph-based biomedical data analytics and AI-driven virtual screening integrated with robotic cloud labs. Through an unbiased, prospective evaluation we show the reliability and robustness of HydraScreen in virtual and high-throughput screening for hit identification in IRAK1. Our platforms and innovative tools can expedite the early stages of drug discovery.

Discovery of the Most Stable Form of an Adenosine Receptor Antagonist through Virtual Polymorph Screening and Targeted Crystallization.

An innovative approach was developed to identify the optimal crystalline form, usually the thermodynamically most stable form. This method involves using virtual polymorph screening and targeted crystallization based on in silico solid-state modeling. By utilizing advanced crystal structure prediction (CSP) technology, the virtual polymorph screening method helps confirm whether the most stable crystalline form has been identified in actual crystallization experiments. If the predicted most stable form is not observed in experiments, predictions based on the method of COnductor like Screening MOdel for Real Solvents (COSMO-RS) are used to highlight solvent systems that can increase the likelihood of experimentally obtaining the desired form through a targeted crystallization process. In this work, such an approach has enabled the rapid discovery of the most stable polymorphic form and the development of a crystallization process of an adenosine receptor antagonist using minimal amounts of the sample within a shortened timeframe. Additionally, it provides a scientific rationale for ensuring the selection of the most stable form in the early stages of drug discovery, thereby reducing risks in future pharmaceutical development.

Use of Nanotechnology in Cosmetics and Cosmeceuticals

Nanotechnology has revolutionized the cosmetic and cosmeceutical industries by enhancing product efficacy and enabling innovative formulations. This review paper delves into the various applications of nanotechnology in skincare and cosmetic products, highlighting its ability to improve the delivery of active ingredients, provide enhanced protection, and increase the bioavailability of beneficial compounds. Nanomaterials such as liposomes, nanoemulsions, nanocapsules, and solid lipid nanoparticles are now extensively utilized for their superior skin penetration capabilities, ensuring targeted delivery and prolonged effects. The paper explores the potential benefits and safety concerns associated with these nano-based cosmetic products. It also provides a comprehensive analysis of recent advancements, regulatory considerations, and future trends in the field, emphasizing the need for further research to fully understand the long-term implications of nanotechnology in cosmetics and cosmeceuticals. This review aims to provide a critical overview for researchers and industry professionals interested in the evolving landscape of nanotechnology in beauty and skincare. Findings Our research highlights the transformative impact of Artificial Intelligence (AI) and Machine Learning (ML) on the drug discovery and development process, particularly in enhancing efficiency and precision from initial screening to clinical trials. The integration of AI/ML has shown significant advancements in early-stage drug discovery, where data-driven algorithms enable rapid identification of potential drug candidates, reducing reliance on traditional, labor-intensive methods. In the drug design and optimization phase, AI-driven predictive models have streamlined the process, minimizing the need for extensive physical testing by accurately simulating drug interactions and predicting possible side effects. Additionally, AI and ML are revolutionizing clinical trials by optimizing trial design, improving patient recruitment and retention, and enhancing real-time data monitoring, leading to faster and more reliable trial outcomes. These technologies also support personalized medicine approaches and have proven essential in reducing both the time and cost associated with bringing new therapies to market. Overall, our findings underscore the critical role of AI and ML in reshaping the pharmaceutical landscape, making drug development faster, more cost-effective, and ultimately, more successful in delivering effective treatments to patients.

GTransCYPs: an improved graph transformer neural network with attention pooling for reliably predicting CYP450 inhibitors

State‑of‑the‑art medical studies proved that predicting CYP450 enzyme inhibitors is beneficial in the early stage of drug discovery. However, accurate machine learning-based (ML) in silico methods for predicting CYP450 inhibitors remains challenging. Here, we introduce GTransCYPs, an improved graph neural network (GNN) with a transformer mechanism for predicting CYP450 inhibitors. This model significantly enhances the discrimination between inhibitors and non-inhibitors for five major CYP450 isozymes: 1A2, 2C9, 2C19, 2D6, and 3A4. GTransCYPs learns information patterns from molecular graphs by aggregating node and edge representations using a transformer. The GTransCYPs model utilizes transformer convolution layers to process features, followed by a global attention-pooling technique to synthesize the graph-level information. This information is then fed through successive linear layers for final output generation. Experimental results demonstrate that the GTransCYPs model achieved high performance, outperforming other state-of-the-art methods in CYP450 prediction.Scientific contributionThe prediction of CYP450 inhibition via computational techniques utilizing biological information has emerged as a cost-effective and highly efficient approach. Here, we presented a deep learning (DL) architecture based on GNN with transformer mechanism and attention pooling (GTransCYPs) to predict CYP450 inhibitors. Four GTransCYPs of different pooling technique were tested on an experimental tasks on the CYP450 prediction problem for the first time. Graph transformer with attention pooling algorithm achieved the best performances. Comparative and ablation experiments provide evidence of the efficacy of our proposed method in predicting CYP450 inhibitors. The source code is publicly available at https://github.com/zonwoo/GTransCYPs.

GCLmf: A Novel Molecular Graph Contrastive Learning Framework Based on Hard Negatives and Application in Toxicity Prediction.

In silico methods for prediction of chemical toxicity can decrease the cost and increase the efficiency in the early stage of drug discovery. However, due to low accessibility of sufficient and reliable toxicity data, constructing robust and accurate prediction models is challenging. Contrastive learning, a type of self-supervised learning, leverages large unlabeled data to obtain more expressive molecular representations, which can boost the prediction performance on downstream tasks. While molecular graph contrastive learning has gathered growing attentions, current models neglect the quality of negative data set. Here, we proposed a self-supervised pretraining deep learning framework named GCLmf. We first utilized molecular fragments that meet specific conditions as hard negative samples to boost the quality of the negative set and thus increase the difficulty of the proxy tasks during pre-training to learn informative representations. GCLmf has shown excellent predictive power on various molecular property benchmarks and demonstrates high performance in 33 toxicity tasks in comparison with multiple baselines. In addition, we further investigated the necessity of introducing hard negatives in model building and the impact of the proportion of hard negatives on the model.

A Cell-Free Kinetic Analysis of Ionizable Lipid Hydrolysis.

The prolonged retention of ionizable lipids within the body limits the repeated dosing of lipid nanoparticles (LNPs) for nucleic acid delivery. While most ionizable lipids are primarily metabolized in the liver via the enzymatic hydrolysis of ester bonds, elimination half-lives can range from several hours to days. The development of compounds that undergo rapid biodegradation remains a major engineering challenge in the absence of standardized biodegradability assessments in the early stages of drug discovery. Here, we analyze and compare the hydrolysis kinetics of well-known ionizable lipids (ALC-0315, DLin-MC3-DMA, LP-01, L319, and SM-102) using optimized cell-free reactions monitored by 1H NMR. Unlike conventional analytical techniques, these NMR-based methods are universal and suitable for high-throughput screening. We demonstrate that enzyme-catalyzed and base hydrolysis reactions can predict whether ionizable lipids undergo fast or slow liver elimination, as our results are in alignment with prior pharmacokinetic studies. Furthermore, we show that the hydrolysis kinetics of ionizable lipids vary by several orders of magnitude depending on steric effects. This study provides a framework to expedite the discovery of rapidly degradable ionizable lipids, with implications for improving the therapeutic index of LNP-based drugs.

ChemAP: predicting drug approval with chemical structures before clinical trial phase by leveraging multi-modal embedding space and knowledge distillation

Recent studies showed that the likelihood of drug approval can be predicted with clinical data and structure information of drug using computational approaches. Predicting the likelihood of drug approval can be innovative and of high impact. However, models that leverage clinical data are applicable only in clinical stages, which is not very practical. Prioritizing drug candidates and early-stage decision-making in the de novo drug development process is crucial in pharmaceutical research to optimize resource allocation. For early-stage decision-making, we need a computational model that uses only chemical structures. This seemingly impossible task may utilize the predictive power with multi-modal features including clinical data. In this work, we introduce ChemAP (Chemical structure-based drug Approval Predictor), a novel deep learning scheme for drug approval prediction in the early-stage drug discovery phase. ChemAP aims to enhance the possibility of early-stage decision-making by enriching semantic knowledge to fill in the gap between multi-modal and single-modal chemical spaces through knowledge distillation techniques. This approach facilitates the effective construction of chemical space solely from chemical structure data, guided by multi-modal knowledge related to efficacy, such as clinical trials and patents of drugs. In this study, ChemAP achieved state-of-the-art performance, outperforming both traditional machine learning and deep learning models in drug approval prediction, with AUROC and AUPRC scores of 0.782 and 0.842 respectively on the drug approval benchmark dataset. Additionally, we demonstrated its generalizability by outperforming baseline models on a recent external dataset, which included drugs from the 2023 FDA-approved list and the 2024 clinical trial failure drug list, achieving AUROC and AUPRC scores of 0.694 and 0.851. These results demonstrate that ChemAP is an effective method in predicting drug approval only with chemical structure information of drug so that decision-making can be done at the early stages of drug development process. To the best of our knowledge, our work is the first of its kind to show that prediction of drug approval is possible only with structure information of drug by defining the chemical space of approved and unapproved drugs using deep learning technology.

Developing Robust Human Liver Microsomal Stability Prediction Models: Leveraging Inter-Species Correlation with Rat Data.

Objectives: Pharmacokinetic issues were the leading cause of drug attrition, accounting for approximately 40% of all cases before the turn of the century. To this end, several high-throughput in vitro assays like microsomal stability have been developed to evaluate the pharmacokinetic profiles of compounds in the early stages of drug discovery. At NCATS, a single-point rat liver microsomal (RLM) stability assay is used as a Tier I assay, while human liver microsomal (HLM) stability is employed as a Tier II assay. We experimentally screened and collected data on over 30,000 compounds for RLM stability and over 7000 compounds for HLM stability. Although HLM stability screening provides valuable insights, the increasing number of hits generated, along with the time- and resource-intensive nature of the assay, highlights the need for alternative strategies. One promising approach is leveraging in silico models trained on these experimental datasets. Methods: We describe the development of an HLM stability prediction model using our in-house HLM stability dataset. Results: Employing both classical machine learning methods and advanced techniques, such as neural networks, we achieved model accuracies exceeding 80%. Moreover, we validated our model using external test sets and found that our models are comparable to some of the best models in literature. Additionally, the strong correlation observed between our RLM and HLM data was further reinforced by the fact that our HLM model performance improved when using RLM stability predictions as an input descriptor. Conclusions: The best model along with a subset of our dataset (PubChem AID: 1963597) has been made publicly accessible on the ADME@NCATS website for the benefit of the greater drug discovery community. To the best of our knowledge, it is the largest open-source model of its kind and the first to leverage cross-species data.

The Development and Application of KinomePro-DL: A Deep Learning Based Online Small Molecule Kinome Selectivity Profiling Prediction Platform.

Characterizing the kinome selectivity profiles of kinase inhibitors is essential in the early stages of novel small-molecule drug discovery. This characterization is critical for interpreting potential adverse events caused by off-target polypharmacology effects and provides unique pharmacological insights for drug repurposing development of existing kinase inhibitor drugs. However, experimental profiling of whole kinome selectivity is still time-consuming and resource-demanding. Here, we report a deep learning classification model using an in-house built data set of inhibitors against 191 well-representative kinases constructed based on a novel strategy by systematically cleaning and integrating six public data sets. This model, a multitask deep neural network, predicts the kinome selectivity profiles of compounds with novel structures. The model demonstrates excellent predictive performance, with auROC, prc-AUC, Accuracy, and Binary_cross_entropy of 0.95, 0.92, 0.90, and 0.37, respectively. It also performs well in a priori testing for inhibitors targeting different categories of proteins from internal compound collections, significantly improving over similar models on data sets from practical application scenarios. Integrated to subsequent machine learning-enhanced virtual screening workflow, novel CDK2 kinase inhibitors with potent kinase inhibitory activity and excellent kinome selectivity profiles are successfully identified. Additionally, we developed a free online web server, KinomePro-DL, to predict the kinome selectivity profiles and kinome-wide polypharmacology effects of small molecules (available on kinomepro-dl.pharmablock.com). Uniquely, our model allows users to quickly fine-tune it with their own training data sets, enhancing both prediction accuracy and robustness.

Application of Ensemble Machine Learning Methods for QSAR Classification of Leukotriene A4 Hydrolase Inhibitors in Drug Discovery

Inflammatory diseases such as asthma, rheumatoid arthritis, and cardiovascular conditions are driven by overproduction of leukotriene B4 (LTB4), a potent inflammatory mediator. Leukotriene A4 hydrolase (LTA4H) plays a critical role in converting leukotriene A4 into LTB4, making it a prime target for drug discovery. Despite ongoing efforts, developing effective LTA4H inhibitors has been challenging due to the complex binding properties of the enzyme and the structural diversity of potential inhibitors. Traditional drug discovery methods, like high-throughput screening (HTS), are often time-consuming and inefficient, prompting the need for more advanced approaches. Quantitative Structure-Activity Relationship (QSAR) modeling, enhanced by ensemble machine learning techniques, provides a promising solution by enabling accurate prediction of compound bioactivity based on molecular descriptors. In this study, six ensemble machine learning methods—AdaBoost, Extra Trees, Gradient Boosting, LightGBM, Random Forest, and XGBoost—were employed to classify LTA4H inhibitors. The dataset, comprising 636 compounds labeled as active or inactive based on pIC50 values, was processed to extract 450 molecular descriptors after feature engineering. The results show that the LightGBM model achieved the highest classification accuracy (83.59%) and Area Under the Curve (AUC) value (0.901), outperforming other models. XGBoost and Random Forest also demonstrated strong performance, with AUC values of 0.890 and 0.895, respectively. The high sensitivity (95.24%) of the XGBoost model highlights its ability to accurately identify active compounds, though it exhibited slightly lower specificity (61.36%), indicating a higher false-positive rate. These findings suggest that ensemble machine learning models, particularly LightGBM, are highly effective in predicting bioactivity, offering valuable tools for early-stage drug discovery. The results indicate that ensemble methods significantly enhance QSAR model accuracy, making them viable for identifying promising LTA4H inhibitors, potentially accelerating the development of anti-inflammatory therapies.

A capability to promote the utilization of human samples for improving the ‍clinical response predictability in the pharmaceutical company

An issue in drug discovery research at pharmaceutical companies is the decline in the probability of market launch, and there is a need to improve the proof-of-concept (POC) acquisition rate by further improving clinical predictability. For this purpose, we need to deepen our understanding of human pathophysiology, and to make efficient of drug discovery research by utilizing human samples and data due to the change in mindset from "animals" to "humans" at the drug discovery research stage. In particular, with the aim of improving the efficiency of drug discovery research by utilizing human samples/data, we have established a human sample utilization support capability, then we are supporting the acquisition of appropriate human samples/data to meet each drug research need with collecting information on organizations that can provide various human samples and accumulating know-how on obtaining human samples/data. In addition, we have built a one-stop support system for surveying human samples/data, negotiating, and contracting with the organizations, obtaining ethical committee approvals, importing samples, and dealing with customs clearance. As results, our researchers can obtain high-quality human samples/data to meet their research needs rapidly/easily. By establishing this capability, the number of the research projects that implement target validation, pharmacological evaluation, BM identification, and patient stratification, etc. using human samples/data from the early stage of drug discovery, has increased, and then, the evidence obtained by using human samples/data contribute to create drug candidates with high probability of clinical effect. We will introduce our activities while showing the support flow that was actually constructed.

Transforming Drug Discovery: The Impact of Artificial Intelligence and Machine Learning from Initial Screening to Clinical Trials

Abstract: The landscape of drug discovery is undergoing a profound transformation driven by the integration of Artificial Intelligence (AI) and Machine Learning (ML) technologies. This paper explores the impact of these advancements on various stages of the drug discovery process, from initial screening to clinical trials. Initially, AI and ML are revolutionizing the early stages of drug discovery by enhancing the efficiency of target identification and lead compound screening. Traditional methods, which often rely on labor-intensive processes and high costs, are being supplemented with AI-driven algorithms that analyze vast datasets to identify potential drug targets and predict the biological activity of compounds with unprecedented accuracy. In the drug design and optimization phase, ML models facilitate the prediction of drug interactions and side effects, thus accelerating the development of safer and more effective therapeutics. Advanced simulations and predictive models reduce the reliance on experimental trials, thereby streamlining the development pipeline. The clinical trials phase also benefits significantly from AI and ML. These technologies improve patient stratification by identifying suitable candidates based on genetic and clinical data, optimizing trial designs, and predicting patient responses to treatment. This not only enhances the efficiency of clinical trials but also increases the likelihood of successful outcome. Findings Our research highlights the transformative impact of Artificial Intelligence (AI) and Machine Learning (ML) on the drug discovery and development process, particularly in enhancing efficiency and precision from initial screening to clinical trials. The integration of AI/ML has shown significant advancements in early-stage drug discovery, where data-driven algorithms enable rapid identification of potential drug candidates, reducing reliance on traditional, labor-intensive methods. In the drug design and optimization phase, AI-driven predictive models have streamlined the process, minimizing the need for extensive physical testing by accurately simulating drug interactions and predicting possible side effects. Additionally, AI and ML are revolutionizing clinical trials by optimizing trial design, improving patient recruitment and retention, and enhancing real-time data monitoring, leading to faster and more reliable trial outcomes. These technologies also support personalized medicine approaches and have proven essential in reducing both the time and cost associated with bringing new therapies to market. Overall, our findings underscore the critical role of AI and ML in reshaping the pharmaceutical landscape, making drug development faster, more cost-effective, and ultimately, more successful in delivering effective treatments to patients.

Accessing Active Fragments for Drug Discovery Utilising Nitroreductase Biocatalysis.

Biocatalysis has played a limited role in the early stages of drug discovery. This is often attributed to the limited substrate scope of enzymes not affording access to vast areas of novel chemical space. Here, we have shown a promiscuous nitroreductase enzyme (NR-55) can be used to produce a panel of functionalised anilines from a diverse panel of aryl nitro starting materials. After screening on analytical scale, we show that sixteen substrates could be scaled to 1 mmol scale, with several poly-functional anilines afforded with ease under the standard conditions. The aniline products were also screened for activity against several cell lines of interest, with modest activity observed for one compound. This study demonstrates the potential for nitroreductase biocatalysis to provide access to functional fragments under benign conditions.

Machine Learning Meets Physics-based Modeling: A Mass-spring System to Predict Protein-ligand Binding Affinity.

Computational assessment of the energetics of protein-ligand complexes is a challenge in the early stages of drug discovery. Previous comparative studies on computational methods to calculate the binding affinity showed that targeted scoring functions outperform universal models. The goal here is to review the application of a simple physics-based model to estimate the binding. The focus is on a mass-spring system developed to predict binding affinity against cyclin-dependent kinase. Publications in PubMed were searched to find mass-spring models to predict binding affinity. Crystal structures of cyclin-dependent kinases found in the protein data bank and two web servers to calculate affinity based on the atomic coordinates were employed. One recent study showed how a simple physics-based scoring function (named Taba) could contribute to the analysis of protein-ligand interactions. Taba methodology outperforms robust physics-based models implemented in docking programs such as AutoDock4 and Molegro Virtual Docker. Predictive metrics of 27 scoring functions and energy terms highlight the superior performance of the Taba scoring function for cyclin- dependent kinase. The recent progress of machine learning methods and the availability of these techniques through free libraries boosted the development of more accurate models to address protein-ligand interactions. Combining a naïve mass-spring system with machine-learning techniques generated a targeted scoring function with superior predictive performance to estimate pKi.

GraphCL-DTA: A Graph Contrastive Learning With Molecular Semantics for Drug-Target Binding Affinity Prediction.

Drug-target binding affinity prediction plays an important role in the early stages of drug discovery, which can infer the strength of interactions between new drugs and new targets. However, the performance of previous computational models is limited by the following drawbacks. The learning of drug representation relies only on supervised data without considering the information in the molecular graph itself. Moreover, most previous studies tended to design complicated representation learning modules, while uniformity used to measure representation quality is ignored. In this study, we propose GraphCL-DTA, a graph contrastive learning with molecular semantics for drug-target binding affinity prediction. This graph contrastive learning framework replaces the dropout-based data augmentation strategy by performing data augmentation in the embedding space, thereby better preserving the semantic information of the molecular graph. A more essential and effective drug representation can be learned through this graph contrastive framework without additional supervised data. Next, we design a new loss function that can be directly used to adjust the uniformity of drug and target representations. By directly optimizing the uniformity of representations, the representation quality of drugs and targets can be improved. The effectiveness of the above innovative elements is verified on two real datasets, KIBA and Davis. Compared with the GraphDTA model, the relative improvement of the GraphCL-DTA model on the two datasets is 2.7% and 4.5%. The graph contrastive learning framework and uniformity function in the GraphCL-DTA model can be embedded into other computational models as independent modules to improve their generalization capability.

Prediction of the binding interactions between rosmarinic acid and cysteinyl leukotriene receptor type 1 by molecular docking and immobilized receptor chromatography.

Drug-protein interaction analysis is still at the center of research efforts to illustrate binding mechanisms and provide valuable information for selecting drug candidates with ideal properties in the early drug discovery stage. We present the prediction of the binding of rosmarinic acid (RA) to cysteinyl leukotriene receptor type1 (CysLTR1) by molecular docking. According to our findings, CysLTR1 is a potential anti-inflammatory target of RA. Under this assumption, we prepared the immobilized CysLTR1 column via a one-step method and characterized the immobilized CysLTR1 by fluorescent and chromatographic analyses. Furthermore, we used the immobilized CysLTR1 column to evaluate the binding interactions between RA and the immobilized receptor. Molecular docking showed that Tyr 249, Phe 174, Thr 280, Pro 177, and Thr 100 are the main sites for RA to interact with CysLTR1. The main forces that drive the findings are hydrogen bonds and hydrophobic interactions. Characterization results show that CysLTR1 is successfully immobilized with high specificity and stability. Almost no non-specific binding is observed on the immobilized CysLTR1 gels. The association constant and the binding sites are calculated to be 7.268 × 105 L mol-1 and 1.237 × 10-8 mol L-1 by injection amount-dependent method. These results, taken together, confirm the potential target of RA on the anti-inflammatory effect. We believe that it can provide valuable reference information on the in-depth exploration of drug-protein interaction mechanisms, and lead compound screening by this method.

Rapid, Accurate, Ranking of Protein-Ligand Binding Affinities with VM2, the Second-Generation Mining Minima Method.

The structure-based technologies most widely used to rank the affinities of candidate small molecule drugs for proteins range from faster but less reliable docking methods to slower but more accurate explicit solvent free energy methods. In recent years, we have advanced another technology, which is called mining minima because it "mines" out the main contributions to the chemical potentials of the free and bound molecular species by identifying and characterizing their main local energy minima. The present study provides systematic benchmarks of the accuracy and computational speed of mining minima, as implemented in the VeraChem Mining Minima Generation 2 (VM2) code, across two well-regarded protein-ligand benchmark data sets, for which there are already benchmark data for docking, free energy, and other computational methods. A core result is that VM2's accuracy approaches that of explicit solvent free energy methods at a far lower computational cost. In finer-grained analyses, we also examine the influence of various run settings, such as the treatment of crystallographic water molecules, on the accuracy, and define the costs in time and dollars of representative runs on Amazon Web Services (AWS) compute instances with various CPU and GPU combinations. We also use the benchmark data to determine the importance of VM2's correction from generalized Born to finite-difference Poisson-Boltzmann results for each energy well and find that this correction affords a remarkably consistent improvement in accuracy at a modest computational cost. The present results establish VM2 as a distinctive technology for early-stage drug discovery, which provides a strong combination of efficiency and predictivity.

CNS Drug Discovery in Academia: Where Basic Research Meets Innovation.

The involvement of academic research in drug discovery is consistently growing. However, academic projects seldom advance to clinical trials. Here, we assess the landscape of drug discovery within the National Centre of Competence in Research (NCCR) TransCure launched by the Swiss National Science Foundation to foster basic research and early-stage drug discovery on membrane transporters. This included transporters in central nervous system (CNS) disorders, which represent a huge unmet medical need. While idea championship, sustainable funding, collaborations between disciplines at the interface of academia and industry are important for translational research, Popperian falsifiability, strong intellectual property and a motivated startup team are key elements for innovation. This is exemplified by the NCCR TransCure spin-off company Synendos Therapeutics, a clinical stage biotech company developing the first selective endocannabinoid reuptake inhibitors (SERIs) as novel treatment for neuropsychiatric disorders. We provide a perspective on the challenges related to entering an uncharted druggable space and bridging the often mentioned "valley of death". The high attrition rate of drug discovery projects in the CNS field within academia is often due to the lack of meaningful animal models that can provide pharmacological proof-of-concept for potentially disruptive technologies at the earliest stages, and the absence of solid intellectual property.

Ligand-Based Compound Activity Prediction via Few-Shot Learning.

Predicting the activities of new compounds against biophysical or phenotypic assays based on the known activities of one or a few existing compounds is a common goal in early stage drug discovery. This problem can be cast as a "few-shot learning" challenge, and prior studies have developed few-shot learning methods to classify compounds as active versus inactive. However, the ability to go beyond classification and rank compounds by expected affinity is more valuable. We describe Few-Shot Compound Activity Prediction (FS-CAP), a novel neural architecture trained on a large bioactivity data set to predict compound activities against an assay outside the training set, based on only the activities of a few known compounds against the same assay. Our model aggregates encodings generated from the known compounds and their activities to capture assay information and uses a separate encoder for the new compound whose activity is to be predicted. The new method provides encouraging results relative to traditional chemical-similarity-based techniques as well as other state-of-the-art few-shot learning methods in tests on a variety of ligand-based drug discovery settings and data sets. The code for FS-CAP is available at https://github.com/Rose-STL-Lab/FS-CAP.

Boosting the Accuracy and Chemical Space Coverage of the Detection of Small Colloidal Aggregating Molecules Using the BAD Molecule Filter.

The ability to conduct effective high throughput screening (HTS) campaigns in drug discovery is often hampered by the detection of false positives in these assays due to small colloidally aggregating molecules (SCAMs). SCAMs can produce artifactual hits in HTS by nonspecific inhibition of the protein target. In this work, we present a new computational prediction tool for detecting SCAMs based on their 2D chemical structure. The tool, called the boosted aggregation detection (BAD) molecule filter, employs decision tree ensemble methods, namely, the CatBoost classifier and the light gradient-boosting machine, to significantly improve the detection of SCAMs. In developing the filter, we explore models trained on individual data sets, a consensus approach using these models, and, third, a merged data set approach, each tailored for specific drug discovery needs. The individual data set method emerged as most effective, achieving 93% sensitivity and 90% specificity, outperforming existing state-of-the-art models by 20 and 5%, respectively. The consensus models offer broader chemical space coverage, exceeding 90% for all testing sets. This feature is an important aspect particularly for early stage medicinal chemistry projects, and provides information on applicability domain. Meanwhile, the merged data set models demonstrated robust performance, with a notable sensitivity of 79% in the comprehensive 10-fold cross-validation test set. A SHAP analysis of model features indicates the importance of hydrophobicity and molecular complexity as primary factors influencing the aggregation propensity. The BAD molecule filter is readily accessible for the public usage on https://molmodlab-aau.com/Tools.html. This filter provides a new, more robust tool for aggregate prediction in the early stages of drug discovery to optimize hit rates and reduce associated testing and validation overheads.