Early Stages Of Drug Development Research Articles

A Simple Machine Learning-Based Quantitative Structure-Activity Relationship Model for Predicting pIC50 Inhibition Values of FLT3 Tyrosine Kinase.

Background/Objectives: Acute myeloid leukemia (AML) presents significant therapeutic challenges, particularly in cases driven by mutations in the FLT3 tyrosine kinase. This study aimed to develop a robust and user-friendly machine learning-based quantitative structure-activity relationship (QSAR) model to predict the inhibitory potency (pIC50 values) of FLT3 inhibitors, addressing the limitations of previous models in dataset size, diversity, and predictive accuracy. Methods: Using a dataset which was 14 times larger than those employed in prior studies (1350 compounds with 1269 molecular descriptors), we trained a random forest regressor, chosen due to its superior predictive performance and resistance to overfitting. Rigorous internal validation via leave-one-out and 10-fold cross-validation yielded Q2 values of 0.926 and 0.922, respectively, while external validation on 270 independent compounds resulted in an R2 value of 0.941 with a standard deviation of 0.237. Results: Key molecular descriptors influencing the inhibitor potency were identified, thereby improving the interpretability of structural requirements. Additionally, a user-friendly computational tool was developed to enable rapid prediction of pIC50 values and facilitate ligand-based virtual screening, leading to the identification of promising FLT3 inhibitors. Conclusions: These results represent a significant advancement in the field of FLT3 inhibitor discovery, offering a reliable, practical, and efficient approach for early-stage drug development, potentially accelerating the creation of targeted therapies for AML.

Chemometric Methods-A Valuable Tool for Investigating the Interactions Between Antifungal Drugs (Including Antifungal Antibiotics) and Food.

Developing antifungal drugs with lower potential for interactions with food may help to optimize treatment and reduce the risk of antimicrobial resistance. Chemometrics uses statistical and mathematical methods to analyze multivariate chemical data, enabling the identification of key correlations and simplifying data interpretation. We used the partial least squares (PLS) approach to explore the correlations between various characteristics of oral antifungal drugs (including antifungal antibiotics) and dietary interventions, aiming to identify patterns that could inform the optimization of antifungal therapy. We analyzed 15 oral antifungal drugs, including azoles (8), antifungal antibiotics (4), antifungal antimetabolites (1), squalene epoxidase inhibitors (1), and glucan synthase inhibitors (1). The input dataset comprised information from published clinical trials, chemical records, and calculations. We constructed PLS models with changes in the pharmacokinetic parameters (∆AUC, area under the curve; ∆Cmax, maximum drug concentration; and ∆Tmax, time to reach maximum drug concentration) after dietary intervention as the response parameters and eight groups of molecular descriptors (M1-M8) as the predictor parameters. We performed separate analyses for the different nutritional interventions. In the final PLS model with food as an intervention, we effectively reduced the dimensionality of the dataset while retaining a substantial percentage of the original information (variance), as significant components explained 69.8% and 17.5% of the predictor and response parameter variances, respectively. The PLS model was significant because its components met the cross-validation criteria. We obtained six significant positive and negative correlations between the descriptors related to atoms and the postprandial ∆Tmax. The PLS method is valuable for investigating interactions between antifungal drugs (including antifungal antibiotics) and food. The correlations obtained can be used in drug modeling to predict interactions with dietary interventions based on the antifungal drug's chemical structure. Incorporating chemometric techniques into the early drug development stages could facilitate the design of antifungal antibiotics and other antifungal agents with optimized absorption in the presence of dietary components.

Antibiotic candidates for Gram-positive bacterial infections induce multidrug resistance.

Several antibiotic candidates are in development against Gram-positive bacterial pathogens, but their long-term utility is unclear. To investigate this issue, we studied the laboratory evolution of resistance to antibiotics that have not yet reached the market. We found that, with the exception of compound SCH79797, antibiotic resistance generally readily evolves in Staphylococcus aureus. Cross-resistance was detected between such candidates and antibiotics currently in clinical use, including vancomycin, daptomycin, and the promising antibiotic candidate teixobactin. These patterns were driven by overlapping molecular mechanisms through mutations in regulatory systems. In particular, teixobactin-resistant bacteria displayed clinically relevant multidrug resistance and retained their virulence in an invertebrate infection model, raising concerns. More generally, we demonstrate that putative resistance mutations against candidate antibiotics are already present in natural bacterial populations. Therefore, antibiotic resistance in nature may evolve readily from the selection of preexisting genetic variants. Our work highlights the importance of predicting future evolution of resistance to antibiotic candidates at an early stage of drug development.

FormulationBCS: A Machine Learning Platform Based on Diverse Molecular Representations for Biopharmaceutical Classification System (BCS) Class Prediction.

The Biopharmaceutics Classification System (BCS) has facilitated biowaivers and played a significant role in enhancing drug regulation and development efficiency. However, the productivity of measuring the key discriminative properties of BCS, solubility and permeability, still requires improvement, limiting high-throughput applications of BCS, which is essential for evaluating drug candidate developability and guiding formulation decisions in the early stages of drug development. In recent years, advancements in machine learning (ML) and molecular characterization have revealed the potential of quantitative structure-performance relationships (QSPR) for rapid and accurate in silico BCS classification. The present study aims to develop a web platform for high-throughput BCS classification based on high-performance ML models. Initially, four data sets of BCS-related molecular properties: log S, log P, log D, and log Papp were curated. Subsequently, 6 ML algorithms or deep learning frameworks were employed to construct models, with diverse molecular representations ranging from one-dimensional molecular fingerprints, descriptors, and molecular graphs to three-dimensional molecular spatial coordinates. By comparing different combinations of molecular representations and learning algorithms, LightGBM exhibited excellent performance in solubility prediction, with an R2 of 0.84; AttentiveFP outperformed others in permeability prediction, with R2 values of 0.96 and 0.76 for log P and log D, respectively; and XGBoost was the most accurate for log Papp prediction, with an R2 of 0.71. When externally validated on a marketed drug BCS category data set, the best-performing models achieved classification accuracies of over 77 and 73% for solubility and permeability, respectively. Finally, the well-trained models were embedded into the first ML-based BCS class prediction web platform (x f), enabling pharmaceutical scientists to quickly determine the BCS category of candidate drugs, which will aid in the high-throughput BCS assessment for candidate drugs during the preformulation stage, thereby promoting reduced risk and enhanced efficiency in drug development and regulation.

A high-throughput micro-scale workflow to quantify molecularly dissolved drug concentrations under solubilizing conditions.

Physiological and artificial solubilizing agents usually enhance apparent solubility of poorly soluble drugs, and in many cases also oral drug exposure. However, exposure may decrease in cases where micellization reduces the molecularly dissolved drug fraction, overriding the solubility advantage. While this information is critical to accurately anticipate the effect of drug micellization on oral absorption, the experimental determination of molecularly dissolved drug concentrations is complex and time consuming. The present study aimed at developing a micro-scale in vitro workflow (comparative micro-scale mass transfer assay, CMMTA) to quantify molecularly dissolved (unbound) drug concentrations in the presence of solubilizing agents. A linear correlation was determined between the cumulative passive permeation of the model drug riluzole (RLZ) and its concentration in the donor buffer solution on a 96-well microtiter sandwich plate (PermeaPlain™). Next, the drug permeation from micellar drug solutions (in fasted and fed state simulated intestinal fluids, FaSSIF and FeSSIF) was measured and the concentration of unbound drug was derived from the aforementioned correlation. The results were validated against established methods to measure free (unbound) drug concentrations, namely equilibrium dialysis and microdialysis. The concentrations of molecularly dissolved RLZ were correctly captured on one single microtiter plate. Both, the standard curve and samples at different solubilizing conditions can be determined simultaneously within a few hours using small quantities of drug substance. Hence, the proposed CMMTA workflow represents a promising screening tool for early-stage drug development.

How organ-on-a-chip is advancing cancer research and oncology - a cancer hallmarks’ perspective

Despite significant improvements in methodologies behind the development of new anticancer therapies, the path from early-stage drug development through preclinical and clinical development pipeline is still arduous. Further refinement of the current preclinical models and the development of complementing alternative techniques that enable more reliable studies are paramount. The emerging organ-on-a-chip (OoC) technologies are physiological-like organ biomimetic systems built on a microfluidic chip, capable of enabling precise control over various physicochemical and biomechanical parameters and helping recreate the natural physiology and mechanical forces that cells experience in the human body. In oncology research, especially, since cancer has been understood to be a dynamic disease featured by complex interactions between cancer cells and their environment, this technology offers added advantage as it is able to provide a dynamic platform to simulate cancer-on-a-chip emulating the biological context of tumor microenvironment (TME); demonstrating progression to metastases to multiorgans; and helping to unravel complex information that other current in vitro methods are otherwise not able to provide. In this contribution, we provide a biological perspective on the recent advances in the field of OoC models in cancer biology studies from a cancer hallmark perspective. We also discuss current prospects and funding opportunities in this space, as well as a possible future outlook from a biology perspective, including major challenges and new opportunities as a way forward to OoC technologies.

HERGBoost: A gradient boosting model for quantitative IC50 prediction of hERG channel blockers

The human ether-a-go-go-related gene (hERG) potassium channel is pivotal in drug discovery due to its susceptibility to blockage by drug candidate molecules, which can cause severe cardiotoxic effects. Consequently, identifying and excluding potential hERG channel blockers at the earliest stages of drug development is crucial. Most traditional machine learning models predict a molecule's cardiotoxicity or non-cardiotoxicity typically at 10 μM, which doesn't account for compounds with low IC50 values that are non-toxic at therapeutic levels due to their high effectiveness at lower concentrations. To address the need for more precise, quantitative predictions, we developed hERGBoost, a cutting-edge machine learning model employing a gradient-boosting algorithm. This model demonstrates superior accuracy in predicting the IC50 of drug candidates. Trained on a specially curated dataset for this study, hERGBoost not only exhibited excellent performance in external validation, achieving an R2 score of 0.394 and a low root mean square error of 0.616 but also significantly outstripped previous models in both qualitative and quantitative assessments. Representing a notable leap forward in the prediction of hERG channel blockers, the hERGBoost model and its datasets are freely available to the drug discovery community on our web server at. http://ssbio.cau.ac.kr/software/hergboost This resource promises to be invaluable in advancing safer pharmaceutical development.

Ab-Amy 2.0: Predicting light chain amyloidogenic risk of therapeutic antibodies based on antibody language model

Therapeutic antibodies have emerged as a promising treatment option for a wide range of diseases. However, the light chain of antibodies can potentially induce amyloidosis, a condition characterized by protein misfolding and aggregation, posing a significant safety concern. Therefore, it is crucial to assess the amyloidogenic risk of therapeutic antibodies during the early stages of drug development. In this study, we introduce AB-Amy 2.0, a new computational model with enhanced performance for assessing the light chain amyloidogenic risk of therapeutic antibodies. By employing pretrained protein language models (PLMs) embeddings, AB-Amy 2.0 achieves higher accuracy in amyloidogenic risk prediction compared with traditional features offering a crucial tool for early-stage identification of antibodies with low aggregation propensity. The AB-Amy 2.0 was trained on antiBERTy embeddings and utilizes the SVM algorithm, resulting in superior performance metrics. On an independent test dataset, the model achieved high sensitivity, specificity, ACC, MCC and AUC of 93.47%, 89.23%, 91.92%, 0.8261 and 0.9739, respectively. These results highlight the effectiveness and robustness of AB-Amy 2.0 in predicting light chain amyloidogenic risk accurately. To facilitate user-friendly access, we have developed an online web server (http://i.uestc.edu.cn/AB-Amy2) and a command line tool (https://github.com/zzyywww/ABAmy2). These resources enable the broader application of this advanced model and promise to enhance the development of safer therapeutic antibodies.

NeuroClick: Advanced Software for Designing Blood-Brain Barrier-Permeable Drugs Using Click Reaction Simulations.

NeuroClick is a software tool designed for the in silico execution of azide-alkyne cycloaddition reactions, commonly known as click reactions. We developed this graphical user interface application to expedite the drug discovery process by generating libraries of 1,2,3-triazole compounds. NeuroClick enables users to input reagent SMILES strings, rapidly generating and screening extensive combinatorial libraries at a pace of 10,000 molecules per minute. The software applies stringent criteria to ensure the relevance and accuracy of the generated compounds, excluding molecules without azide groups or those with multiple reactive functional groups to maintain dataset integrity. NeuroClick incorporates advanced filtering options based on Lipinski's rule of five and blood-brain barrier (BBB) permeability predictors, allowing researchers to identify drug-like molecules with potential central nervous system activity. The software's high-throughput and user-friendly interface significantly enhance the efficiency of early-stage drug development by facilitating the exploration of vast chemical spaces and identifying promising lead compounds for further development. This article provides comprehensive guidance on the installation, usage, and features of NeuroClick, ensuring that users can leverage its full potential in their research endeavors. © 2024 Wiley Periodicals LLC. Basic Protocol: NeuroClick workflow for generating BBB-permeating drugs.

Boronic Acid-Based Glucose Detection for Liver Three-Dimensional Cellular Models.

Liver 3D cell models are regularly employed as a screening platform for predicting the metabolic safety of drugs, by monitoring the physiological responses of the spheroids, through the measurement of relevant markers of normal liver physiology, notably glucose. Measuring glucose levels within the spheroids and their surroundings provides insight into the metabolic homeostasis of liver cells and may be employed as an indication of potential drug-induced toxicity. Several ortho-aminomethyl phenylboronic acid (PDBA) glucose sensors have been developed. Most recently, Mc-CDBA ((((((2-(methoxycarbonyl)anthracene-9,10-diyl)bis(methylene)) bis(methylazanediyl))bis(methylene))bis(4-cyano-2,1-phenylene))diboronic acid) was reported. Although Mc-CDBA exhibits good water solubility and sensitivity toward glucose, its ability for intra- and extracellular glucose monitoring in spheroids has not been determined. Here, we present a set of Mc-CDBA derivatives: carboxylic (BA) and amide (BA 5)-based Mc-CDBA sensors for extra- and intracellular glucose monitoring, respectively. Both sensors exhibit superior spectroscopic features. BA 5 showed selective intracellular accumulation in liver spheroids and exhibited more than 3-fold higher basal fluorescence sensitivity compared to Mc-CDBA. These observations led to the development of an extracellular hydrogel-embedded sensor (HG-BA 21) to monitor extracellular glucose levels under persistent solution flow mimicking physiological conditions. We have therefore demonstrated that the sensors developed by our team are suitable for a variety of assays, notably with liver spheroids, and provide powerful new tools for organ-on-a-chip applications predicting drug-induced liver injury in the early stages of drug development.

Quantitative Prediction of Drug-Drug Interactions Caused by CYP3A Induction Using Endogenous Biomarker 4β-Hydroxycholesterol.

Evaluation of the CYP3A induction risk is important in early drug development stages. This study focused on 4β-hydroxycholesterol (4β-HC) as an endogenous biomarker of drug-drug interactions (DDIs) caused by CYP3A induction. We investigated a new approach using 4β-HC for quantitative prediction of DDIs caused by CYP3A induction based on the mechanistic static pharmacokinetic (MSPK) model. The induction ratio, i.e., the ratio of plasma 4β-HC or 4β-HC/cholesterol (4β-HC/C) with and without a coadministered CYP3A inducer, and the ratio of the area under the plasma concentration-time curve (AUCR), i.e., the ratio of the AUC of plasma CYP3A substrate drugs with and without a coadministered CYP3A inducer, were collected. The scaling factor (d) in the MSPK model was calculated from the induction ratio of 4β-HC or 4β-HC/C based on the systemic term in the MSPK model. The AUCR of 18 CYP3A substrates with and without coadministration of seven CYP3A inducers were then predicted by substituting the calculated d value into the MSPK model. This approach showed that approximately 84% of the predicted AUCR values were within a twofold range of the observed values, showing that this approach can be a good tool to quantitatively predict DDIs caused by CYP3A induction. SIGNIFICANCE STATEMENT: A concise approach to predict drug interactions with adequate accuracy is preferable in the early drug development stage. In this study, a new approach using 4β-hydroxycholesterol for quantitative prediction of drug-drug interactions caused by CYP3A induction was investigated. The predictability was verified using seven CYP3A inducers and 18 substrates.

WaSPred: A reliable AI-based water solubility predictor for small molecules

A rapid and reliable evaluation of the aqueous solubility of small molecules is a hot topic for the scientific community and represents a field of particular interest in drug discovery. In fact, aqueous solubility significantly impacts various aspects that collectively influence a drug’s overall pharmacokinetics, including absorption, distribution and metabolism. For this reason, in silico approaches that provide fast and cost-effective solubility predictions, can serve as invaluable tools in the early stages of drug development. Although additional molecular features should be considered, accurate solubility predictions can help medicinal chemists rationally planning the synthesis of compounds more likely to exhibit desirable pharmacokinetic properties and in selecting the most promising candidates for further biological testing (e.g., cellular assays) from an initial pool of hit compounds with detected preliminary bioactivity. In this context, we herein report the development and evaluation of WaSPred, our AI-based water solubility predictor for small molecules. WaSPred not only showed high reliability in water solubility predictions performed on structurally heterogeneous compounds, belonging to multiple external datasets, but also demonstrated superior performance compared to a set of other commonly used water solubility predictors, thus confirming its state-of the-art robustness and its usefulness as an in silico approach for water solubility evaluations..

In Vitro Evaluation of Pharmacokinetic Properties of Selected Dual COX-2 and 5-LOX Inhibitors.

Evaluation of pharmacokinetic properties is a significant step at the early stages of drug development. In this study, an in vitro evaluation of the pharmacokinetic properties of five newly synthesized compounds was performed. These compounds belong to N-hydroxyurea and hydroxamic acid derivatives and analogs of NSAIDs indomethacin, flurbiprofen, diclofenac, ibuprofen, and naproxen (compounds 1, 2, 3, 11, and 12, respectively) with dual COX-2 and 5-LOX inhibitory activity. Two in vitro methods (biopartitioning micellar chromatography (BMC) and PAMPA) were used to evaluate passive gastrointestinal absorption, while high-performance affinity chromatography (HPAC) and differential pulse voltammetry (DPV) were used to evaluate binding to human serum albumin (HSA). The introduction of N-hydroxyurea and hydroxamic acid groups into the structure of NSAIDs decreases both expected passive gastrointestinal absorption (BMC k values were from 3.02 to 9.50, while for NSAIDs were from 5.29 to 13.36; PAMPA -logPe values were between 3.81 and 4.76, while for NSAIDs were ≤3.46) and HSA binding (HPAC logk values were from 2.03 to 9.54, while for NSAIDs were ≥11.03; DPV peak potential shifts were between 7 and 34, while for NSAIDs were ≥54). Structural modifications of all tested compounds that increase lipophilicity could be considered to enhance their passive gastrointestinal absorption. Considering lower expected HSA binding and higher lipophilicity of tested compounds compared to corresponding NSAIDs, it can be expected that the volume of distribution of compounds 1, 2, 3, 11, and 12 will be higher. Reduced HSA binding may also decrease interactions with other drugs in comparison to corresponding NSAIDs. All tested compounds showed significant microsomal instability (25.07-58.44% decrease in concentration) in comparison to indomethacin (14.47%) and diclofenac (20.99%).

Widespread loss-of-function mutations implicating preexisting resistance to new or repurposed anti-tuberculosis drugs

BackgroundFive New or Repurposed Drugs (NRDs) were approved in the last decade for treatment of multi-drug resistant tuberculosis: bedaquiline, clofazimine, linezolid, delamanid, and pretomanid. Unfortunately, resistance to these drugs emerged faster than anticipated, potentially due to preexisting resistance in naïve strains. Previous investigations into the rapid emergence have mostly included short variants. For the first time, we utilize de novo-assembled genomes, and systematically include Structural Variations (SV) and heterogeneity to comprehensively study this rapid emergence. We show high prevalence of preexisting resistance, identify novel markers of resistance, and lay the foundation for preventing preexisting resistance in future drug development. MethodsFirst, a systematic literature review revealed 313 NRD resistance variants in 13 genes. Next, 409 globally diverse clinical isolates collected prior to the drugs’ programmatic use (308 were multidrug resistant, 106 had de novo assembled genomes) were utilized to study the 13 genes comprehensively for conventional, structural, and heterogeneous variants. FindingsWe identified 5 previously reported and 67 novel putative NRD resistance variants. These variants were 2 promoter mutations (in 8/409 isolates), 13 frameshifts (21/409), 6 SVs (9/409), 35 heterogeneous frameshifts (32/409) and 11 heterogeneous SVs (12/106). Delamanid and pretomanid resistance mutations were most prevalent (48/409), while linezolid resistance mutations were least prevalent (8/409). InterpretationPreexisting mutations implicated in resistance to at least one NRD was highly prevalent (85/409, 21 %). This was mostly caused by loss-of-function mutations in genes responsible for prodrug activation and efflux pump regulation. These preexisting mutations may have emerged through a bet-hedging strategy, or through cross-resistance with non-tuberculosis drugs such as metronidazole. Future drugs that could be resisted through loss-of-function in non-essential genes may suffer from preexisting resistance. The methods used here for comprehensive preexisting resistance assessment (especially SVs and heterogeneity) may mitigate this risk during early-stage drug development.

Temperature mapping of milling by dual centrifugation: A systematic investigation

Using low quantities of drug compounds is often favorable in the early stages of drug development, especially for what require a large screening investigation to define the final formulation composition, such as nano- and microsuspensions. For that reason, the dual centrifugation approach has in the recent years been used due to its reproducible and fast-milling capacity with 40 samples in 2 mL vials simultaneously without the addition of cooling breaks due to a built-in cooling system. Nonetheless, heat can be dissipated into the samples during high-intensity milling, resulting in increased sample temperatures that potentially can affect thermolabile compounds and potential influence the obtained suspensions in the screening experiments if the used stabilizer has temperature dependent variations in the performance. Hence, a systematic investigation of the influence of different process parameters on the heat dissipation in samples during milling by the dual centrifugation approach was performed in the present study. It was found that the milling speed had the highest impact on the final sample temperature, but also other parameters, such as the bead loading, bead size, and placement in the centrifuge during milling had significantly influenced the final mean temperature of the milling media. Higher temperatures were obtained with higher bead loadings, i.e., 3000 mg milling beads/mL and milling speeds (1500 rpm), and when smaller milling beads, i.e., 0.1 mm, were used during production. The study further showed that higher temperatures were measured for samples located on the bottom disk during milling, and also when located on the outer placement on the sample disk. Upscale investigations showed immensely increased sample temperatures (almost up to boiling point) when samples were prepared under similar formulation parameters and milling speed as small-volume vials. Furthermore, the study indicated that the addition of drug compounds during suspension preparation decreased the final sample temperature compared to samples that only contained purified water due to energy absorption of the drug compound.

Prediction of Cytochrome P450 Substrates Using the Explainable Multitask Deep Learning Models.

Cytochromes P450 (P450s or CYPs) are the most important phase I metabolic enzymes in the human body and are responsible for metabolizing ∼75% of the clinically used drugs. P450-mediated metabolism is also closely associated with the formation of toxic metabolites and drug-drug interactions. Therefore, it is of high importance to predict if a compound is the substrate of a given P450 in the early stage of drug development. In this study, we built the multitask learning models to simultaneously predict the substrates of five major drug-metabolizing P450 enzymes, namely, CYP3A4, 2C9, 2C19, 2D6, and 1A2, based on the collected substrate data sets. Compared to the single-task model and conventional machine learning models, the multitask fingerprints and graph neural networks model achieved superior performance with the average AUC values of 90.8% on the test set. Notably, the multitask model demonstrated its good performance on the small amount of substrate data sets such as CYP1A2, 2C9, and 2C19. In addition, the Shapley additive explanation and the attention mechanism were used to reveal specific substructures associated with P450 substrates, which were further confirmed and complemented by the substructure mining tool and the literature.

Development and characterization of dendritic cell internalization and activation assays contributing to the immunogenicity risk evaluation of biotherapeutics.

Immunogenicity refers to the ability of a substance, such as a therapeutic drug, to elicit an immune response. While beneficial in vaccine development, undesirable immunogenicity can compromise the safety and efficacy of therapeutic proteins by inducing anti-drug antibodies (ADAs). These ADAs can reduce drug bioavailability and alter pharmacokinetics, necessitating comprehensive immunogenicity risk assessments starting at early stages of drug development. Given the complexity of immunogenicity, an integrated approach is essential, as no single assay can universally recapitulate the immune response leading to the formation of anti-drug antibodies. To better understand the Dendritic Cell (DC) contribution to immunogenicity, we developed two flow cytometry-based assays: the DC internalization assay and the DC activation assay. Monocyte-derived dendritic cells (moDCs) were generated from peripheral blood mononuclear cells (PBMCs) and differentiated over a five-day period. The internalization assay measured the accumulation rate of therapeutic antibodies within moDCs, while the activation assay assessed the expression of DC activation markers such as CD40, CD80, CD86, CD83, and DC-SIGN (CD209). To characterize these two assays further, we used a set of marketed therapeutic antibodies. The study highlights that moDCs differentiated for 5 days from freshly isolated monocytes were more prone to respond to external stimuli. The internalization assay has been shown to be highly sensitive to the molecule tested, allowing the use of only 4 donors to detect small but significant differences. We also demonstrated that therapeutic antibodies were efficiently taken up by moDCs, with a strong correlation with their peptide presentation on MHC-II. On the other hand, by monitoring DC activation through a limited set of activation markers including CD40, CD83, and DC-SIGN, the DC activation assay has the potential to compare a series of compounds. These two assays provide a more comprehensive understanding of DC function in the context of immunogenicity, highlighting the importance of both internalization and activation processes in ADA development. The DC internalization and activation assays described here address key gaps in existing immunogenicity assessment methods by providing specific and reliable measures of DC function. The assays enhance our ability to pre-clinically evaluate the immunogenic potential of biotherapeutics, thereby improving their safety and efficacy. Future work should focus on further validating these assays and integrating them into a holistic immunogenicity risk assessment framework.

Graph Association Analyses for Early Drug Discovery

We demonstrate MedHunter, a system for assisting the early stage of drug development. MedHunter builds a biomedical knowledge graph DDKG by integrating data from eleven biochemical libraries and data banks, and aligning entities from different data sources by means of heterogeneous entity resolution. It identifies drug-disease associations and protein-protein interactions in DDKG by employing graph association rules (GARs). GARs use graph patterns to extract relevant entities and embed ML models as predicates. MedHunter discovers GARs from DDKG and incrementally enriches DDKG with external data; it cleans DDKG with a special form of GARs. We demonstrate MedHunter for its (a) interfaces, (b) data enrichment/cleaning, and (c) applications in target identification, drug-drug interaction and protein-protein interaction.

Natural Products Have Increased Rates of Clinical Trial Success throughout the Drug Development Process.

Natural products (NPs) or their derivatives represent a large proportion of drugs that successfully progress through clinical trials to approval. This study explores the presence of NPs in both early- and late-stage drug discovery to determine their success rate, and the factors or features of natural products that contribute to such success. As a proxy for early drug development stages, we analyzed patent applications over several decades, finding a consistent proportion of NP, NP-derived, and synthetic-compound-based patent documents, with the latter group outnumbering NP and NP-derived ones (approximately 77% vs 23%). We next assessed clinical trial data, where we observed a steady increase in NP and NP-derived compounds from clinical trial phases I to III (from approximately 35% in phase I to 45% in phase III), with an inverse trend observed in synthetics (from approximately 65% in phase I to 55% in phase III). Finally, in vitro and in silico toxicity studies revealed that NPs and their derivatives were less toxic alternatives to their synthetic counterparts. These discoveries offer valuable insights for successful NP-based drug development, highlighting the potential benefits of prioritizing NPs and their derivatives as starting points.

CavitOmiX Drug Discovery: Engineering Antivirals with Enhanced Spectrum and Reduced Side Effects for Arboviral Diseases.

Advancing climate change increases the risk of future infectious disease outbreaks, particularly of zoonotic diseases, by affecting the abundance and spread of viral vectors. Concerningly, there are currently no approved drugs for some relevant diseases, such as the arboviral diseases chikungunya, dengue or zika. The development of novel inhibitors takes 10-15 years to reach the market and faces critical challenges in preclinical and clinical trials, with approximately 30% of trials failing due to side effects. As an early response to emerging infectious diseases, CavitOmiX allows for a rapid computational screening of databases containing 3D point-clouds representing binding sites of approved drugs to identify candidates for off-label use. This process, known as drug repurposing, reduces the time and cost of regulatory approval. Here, we present potential approved drug candidates for off-label use, targeting the ADP-ribose binding site of Alphavirus chikungunya non-structural protein 3. Additionally, we demonstrate a novel in silico drug design approach, considering potential side effects at the earliest stages of drug development. We use a genetic algorithm to iteratively refine potential inhibitors for (i) reduced off-target activity and (ii) improved binding to different viral variants or across related viral species, to provide broad-spectrum and safe antivirals for the future.