ACFIS 2.0: an improved web-server for fragment-based drug discovery via a dynamic screening strategy.

Drug discovery, which plays a vital role in maintaining human health, is a persistent challenge. Fragment-based drug discovery (FBDD) is one of the strategies for the discovery of novel candidate compounds. Computational tools in FBDD could help to identify potential drug leads in a cost-efficient and time-saving manner. The Auto Core Fragment in silico Screening (ACFIS) server is a well-established and effective online tool for FBDD. However, the accurate prediction of protein-fragment binding mode and affinity is still a major challenge for FBDD due to weak binding affinity. Here, we present an updated version (ACFIS 2.0), that incorporates a dynamic fragment growing strategy to consider protein flexibility. The major improvements of ACFIS 2.0 include (i) increased accuracy of hit compound identification (from 75.4% to 88.5% using the same test set), (ii) improved rationality of the protein-fragment binding mode, (iii) increased structural diversity due to expanded fragment libraries and (iv) inclusion of more comprehensive functionality for predicting molecular properties. Three successful cases of drug lead discovery using ACFIS 2.0 are described, including drugs leads to treat Parkinson's disease, cancer, and major depressive disorder. These cases demonstrate the utility of this web-based server. ACFIS 2.0 is freely available at http://chemyang.ccnu.edu.cn/ccb/server/ACFIS2/.

Computational approaches streamlining drug discovery.

Computer-aided drug discovery has been around for decades, although the past few years have seen a tectonic shift towards embracing computational technologies in both academia and pharma. This shift is largely defined by the flood of data on ligand properties and binding to therapeutic targets and their 3D structures, abundant computing capacities and the advent of on-demand virtual libraries of drug-like small molecules in their billions. Taking full advantage of these resources requires fast computational methods for effective ligand screening. This includes structure-based virtual screening of gigascale chemical spaces, further facilitated by fast iterative screening approaches. Highly synergistic are developments in deep learning predictions of ligand properties and target activities in lieu of receptor structure. Here we review recent advances in ligand discovery technologies, their potential for reshaping the whole process of drug discovery and development, as well as the challenges they encounter. We also discuss how the rapid identification of highly diverse, potent, target-selective and drug-like ligands to protein targets can democratize the drug discovery process, presenting new opportunities for the cost-effective development of safer and more effective small-molecule treatments.

Evaluation of the utility of the Beta Human Liver Emulation System (BHLES) for CFSAN's regulatory toxicology program.

Microphysiological systems (MPS), such as organ-on-a-chip platforms, are an emerging alternative model that may be useful for predicting human physiology and/or toxicity. Due to the interest in these platforms, the Center for Food Safety and Applied Nutrition partnered with Emulate to evaluate the utility of the Beta Human Liver Emulation System (BHLES) for its regulatory science program. Using known hepatotoxic compounds (usnic acid, benzbromarone, tamoxifen, and acetaminophen) and compounds that have no reported human cases of liver toxicity (dimethyl sulfoxide, theophylline, and aminohippurate) the platform's performance was evaluated. Chemical toxicity was assessed by albumin secretion, urea and LDH release, nuclei number, mitochondrial membrane potential, and apoptosis. System/platform performance was evaluated in terms of sensitivity and specificity, power, and variability and repeatability. Chemical interactions with the Chip material were also assessed. Preliminary findings suggested that for the model test compounds selected, the BHLES accurately predicted toxicity, demonstrated high sensitivity and specificity, high power, and low variability. However, some compounds interacted with the Chip material indicating variable exposure levels that should be accounted for when planning experimentation. The details of the evaluation are presented herein.

Identification of in vitro angiotensin-converting enzyme and dipeptidyl peptidase IV inhibitory peptides from draft beer by virtual screening and molecular docking.

BACKGROUND: Hypertension and diabetes are two kinds of senile diseases which often occur simultaneously. The commonly used drugs in clinic may produce certain side effects. Food-derived polypeptide is a kind of polypeptide with great development potential, which has many functions of regulating human physiological function. Beer is rich in nutrition but there are few researches on bioactive peptides in beer. RESULTS: In this study, a rapid virtual screening method was established to obtain bioactive peptides from Tsingtao draft beer. The peptide sequence was analyzed by ultra-performance liquid chromatography-quadrupole-Orbitrap-tandem mass spectrometry (UPLC-Q-Orbitrap-MS2 ), and 50 peptides were identified. Eight peptides with potential biological activities were screened by using Peptide Ranker software and previous literature references. On the basis of absorption prediction, toxicity prediction, and molecular docking analysis, LNFDPNR and LPQQQAQFK were finally confirmed. The molecular docking results showed that two peptides could bind angiotensin-converting enzyme (ACE) and dipeptidyl peptidase IV (DPP-IV) tightly by hydrogen bonding and hydrophobic interaction. The in vitro activity evaluation results showed that two peptides had obvious ACE and DPP-IV inhibitory activity. CONCLUSION: This study established a method for rapidly screening bioactive peptides from Tsingtao draft beer, screened two ACE and DPP-IV inhibitory peptides in beer and analyzed their active action mechanism. This article may have great theoretical significance and practical value to further explore the health function of beer. © 2021 Society of Chemical Industry.

Analysis of Static Molecular Gradients in a High-Throughput Drug Screening Microfluidic Assay.

In this study, we thoroughly analyzed molecular gradient generation, its stability over time, and linearity in our high-throughput drug screening microfluidic assay (HTS). These parameters greatly affect the precision and accuracy of the device's analytical protocol. As part of the research, we developed a mathematical model of dependence of the concentration profile on the initial concentrations of active substances in reservoirs and the number of tilts, as well as the dependence of the active substance concentration profiles in the culture chambers on the concentration profile of the reference dye in the indicator chamber. The mean concentration prediction error of the proposed equations ranged from 1.4% to 2.4% for the optimized parameters of the procedure and did not increase with the incubation time. The concentration profile linearity index, Pearson's correlation coefficient reached -0.997 for 25 device tilts. The observed time stability of the profiles was very good. The mean difference between the concentration profile after 5 days of incubation and the baseline profile was only 7.0%. The newly created mathematical relationships became part of the new HTS biochip operating protocols, which are detailed in the article.

Live-imaging rate-of-kill compound profiling for Chagas disease drug discovery with a new automated high-content assay.

Chagas disease, caused by the protozoan intracellular parasite Trypanosoma cruzi, is a highly neglected tropical disease, causing significant morbidity and mortality in central and south America. Current treatments are inadequate, and recent clinical trials of drugs inhibiting CYP51 have failed, exposing a lack of understanding of how to translate laboratory findings to the clinic. Following these failures many new model systems have been developed, both in vitro and in vivo, that provide improved understanding of the causes for clinical trial failures. Amongst these are in vitro rate-of-kill (RoK) assays that reveal how fast compounds kill intracellular parasites. Such assays have shown clear distinctions between the compounds that failed in clinical trials and the standard of care. However, the published RoK assays have some key drawbacks, including low time-resolution and inability to track the same cell population over time. Here, we present a new, live-imaging RoK assay for intracellular T. cruzi that overcomes these issues. We show that the assay is highly reproducible and report high time-resolution RoK data for key clinical compounds as well as new chemical entities. The data generated by this assay allow fast acting compounds to be prioritised for progression, the fate of individual parasites to be tracked, shifts of mode-of-action within series to be monitored, better PKPD modelling and selection of suitable partners for combination therapy.

Laboratory evaluation of twelve portable devices for medicine quality screening.

BACKGROUND: Post-market surveillance is a key regulatory function to prevent substandard and falsified (SF) medicines from being consumed by patients. Field deployable technologies offer the potential for rapid objective screening for SF medicines. METHODS AND FINDINGS: We evaluated twelve devices: three near infrared spectrometers (MicroPHAZIR RX, NIR-S-G1, Neospectra 2.5), two Raman spectrometers (Progeny, TruScan RM), one mid-infrared spectrometer (4500a), one disposable colorimetric assay (Paper Analytical Devices, PAD), one disposable immunoassay (Rapid Diagnostic Test, RDT), one portable liquid chromatograph (C-Vue), one microfluidic system (PharmaChk), one mass spectrometer (QDa), and one thin layer chromatography kit (GPHF-Minilab). Each device was tested with a series of field collected medicines (FCM) along with simulated medicines (SIM) formulated in a laboratory. The FCM and SIM ranged from samples with good quality active pharmaceutical ingredient (API) concentrations, reduced concentrations of API (80% and 50% of the API), no API, and the wrong API. All the devices had high sensitivities (91.5 to 100.0%) detecting medicines with no API or the wrong API. However, the sensitivities of each device towards samples with 50% and 80% API varied greatly, from 0% to 100%. The infrared and Raman spectrometers had variable sensitivities for detecting samples with 50% and 80% API (from 5.6% to 50.0%). The devices with the ability to quantitate API (C-Vue, PharmaChk, QDa) had sensitivities ranging from 91.7% to 100% to detect all poor quality samples. The specificity was lower for the quantitative C-Vue, PharmaChk, & QDa (50.0% to 91.7%) than for all the other devices in this study (95.5% to 100%). CONCLUSIONS: The twelve devices evaluated could detect medicines with the wrong or none of the APIs, consistent with falsified medicines, with high accuracy. However, API quantitation to detect formulations similar to those commonly found in substandards proved more difficult, requiring further technological innovation.

An integrated biomimetic array chip for establishment of collagen-based 3D primary human hepatocyte model for prediction of clinical drug-induced liver injury.

Drug-induced liver injury (DILI) is a leading cause of therapy failure in the clinic and also contributes much to acute liver failure cases. Investigations of predictive sensitivity in animal models have limitations due to interspecies differences. Previously reported in vitro models of liver injury based on primary human hepatocytes (PHHs) cannot meet the requirements of high physiological fidelity, low cost, simple operation, and high throughput with improved sensitivity. Herein, we developed an integrated biomimetic array chip (iBAC) for establishing extracellular matrix (ECM)-based models. A collagen-based 3D PHH model was constructed on the iBAC as a case for the prediction of clinical DILI at throughput. The iBAC has a three-layer structure with a core component of 3D implanting holes. At an initial cell seeding numbers of 5000-10,000, the collagen-based 3D PHH model was optimized with improved and stabilized liver functionality, including cell viability, albumin, and urea production. Moreover, basal activities of most metabolic enzymes on the iBAC were maintained for at least 12 days. Next, a small-scale hepatotoxicity screening indicated that the 3D PHH model on the iBAC was more sensitive for predicting hepatotoxicity than the 2D PHH model on the plate. Finally, a large-scale screening of liver toxicity using 122 clinical drugs further demonstrated that the collagen-based 3D PHH model on the iBAC had superior predictive sensitivity compared to all previously reported in vitro models. These results indicated the importance of 3D collagen for liver physiological functionality and hepatotoxicity prediction. We anticipant it being a promising tool for risk assessment of drug-induced hepatotoxicity with a widespread acceptance in drug industry.

Evaluation of Normothermic Machine Perfusion of Porcine Livers as a Novel Preclinical Model to Predict Biliary Clearance and Transporter-Mediated Drug-Drug Interactions Using Statins.

There is a lack of translational preclinical models that can predict hepatic handling of drugs. In this study, we aimed to evaluate the applicability of normothermic machine perfusion (NMP) of porcine livers as a novel ex vivo model to predict hepatic clearance, biliary excretion, and plasma exposure of drugs. For this evaluation, we dosed atorvastatin, pitavastatin, and rosuvastatin as model drugs to porcine livers and studied the effect of common drug-drug interactions (DDIs) on these processes. After 120 minutes of perfusion, 0.104 mg atorvastatin (n = 3), 0.140 mg pitavastatin (n = 5), or 1.4 mg rosuvastatin (n = 4) was administered to the portal vein, which was followed 120 minutes later by a second bolus of the statin coadministered with OATP perpetrator drug rifampicin (67.7 mg). After the first dose, all statins were rapidly cleared from the circulation (hepatic extraction ratio > 0.7) and excreted into the bile. Presence of human-specific atorvastatin metabolites confirmed the metabolic capacity of porcine livers. The predicted biliary clearance of rosuvastatin was found to be closer to the observed biliary clearance. A rank order of the DDI between the various systems upon coadministration with rifampicin could be observed: atorvastatin (AUC ratio 7.2) > rosuvastatin (AUC ratio 3.1) > pitavastatin (AUC ratio 2.6), which is in good agreement with the clinical DDI data. The results from this study demonstrated the applicability of using NMP of porcine livers as a novel preclinical model to study OATP-mediated DDI and its effect on hepatic clearance, biliary excretion, and plasma profile of drugs. SIGNIFICANCE STATEMENT: This study evaluated the use of normothermic machine perfusion (NMP) of porcine livers as a novel preclinical model to study hepatic clearance, biliary excretion, plasma (metabolite) profile of statins, and OATP-mediated DDI. Results showed that NMP of porcine livers is a reliable model to study OATP-mediated DDI. Overall, the rank order of DDI severity indicated in these experiments is in good agreement with clinical data, indicating the potential importance of this new ex vivo model in early drug discovery.

Drug Evaluation Based on a Multi-Channel Cell Chip with a Horizontal Co-Culture.

We developed a multi-channel cell chip containing a three-dimensional (3D) scaffold for horizontal co-culture and drug toxicity screening in multi-organ culture (human glioblastoma, cervical cancer, normal liver cells, and normal lung cells). The polydimethylsiloxane (PDMS) multi-channel cell chip (PMCCC) was based on fused deposition modeling (FDM) technology. The architecture of the PMCCC was an open-type cell chip and did not require a pump or syringe. We investigated cell proliferation and cytotoxicity by conducting 3-(4,5-dimethylthiazol-2-yl)-2,5-dphenyltetrazolium bromide (MTT) and lactate dehydrogenase (LDH) assays and analysis of oleanolic acid (OA)-treated multi-channel cell chips. The results of the MTT and LDH assays showed that OA treatment in the multi-channel cell chip of four cell lines enhanced chemoresistance of cells compared with that in the 2D culture. Furthermore, we demonstrated the feasibility of the application of our multi-channel cell chip in various analysis methods through Annexin V-fluorescein isothiocyanate/propidium iodide staining, which is not used for conventional cell chips. Taken together, the results demonstrated that the PMCCC may be used as a new 3D platform because it enables simultaneous drug screening in multiple cells by single point injection and allows analysis of various biological processes.

Multi-Electrode Array of Sensory Neurons as an In Vitro Platform to Identify the Nociceptive Response to Pharmaceutical Buffer Systems of Injectable Biologics.

PURPOSE: Pharmaceutical buffer systems, especially for injectable biologics such as monoclonal antibodies, are an important component of successful FDA-approved medications. Clinical studies indicate that buffer components may be contributing factors for increased injection site pain. METHODS: To determine the potential nociceptive effects of clinically relevant buffer systems, we developed an in vitro multi-electrode array (MEA) based recording system of rodent dorsal root ganglia (DRG) sensory neuron cell culture. This system monitors sensory neuron activity/firing as a surrogate of nociception when challenged with buffer components used in formulating monoclonal antibodies and other injectable biologics. RESULTS: We show that citrate salt and citrate mannitol buffer systems cause an increase in mean firing rate, burst frequency, and burst duration in DRG sensory neurons, unlike histidine or saline buffer systems at the same pH value. Lowering the concentration of citrate leads to a lower firing intensity of DRG sensory neurons. CONCLUSION: Increased activity/firing of DRG sensory neurons has been suggested as a key feature underlying nociception. Our results support the utility of an in vitro MEA assay with cultured DRG sensory neurons to probe the nociceptive potential of clinically relevant buffer components used in injectable biologics.

Assessment of the Effects of Drugs on Mitochondrial Respiration.

Mitochondria are targets of newly synthesized drugs and being tested for the treatment of various diseases caused or accompanied by disruption of cellular bioenergetics. In drug development, it is necessary to test for drug-induced changes in mitochondrial enzyme activity that may be related to therapeutic or adverse drug effects. Measurement of drug effect on mitochondrial oxygen consumption kinetics and/or protective effects of drugs against calcium-induced inhibition of the mitochondrial respiration can be used for the study mitochondrial toxicity and neuroprotective effects of drugs. Supposing that the drug-induced inhibition of the mitochondrial respiratory rate and/or individual mitochondrial complexes is associated with adverse drug effects, the effects of drugs on mitochondrial respiration in isolated mitochondria allow selection of novel molecules that are relatively safe for mitochondrial toxicity.

Drug Transcellular Transport Assay Using a High Porosity Honeycomb Film.

In vitro transport studies across cells grown on culture inserts are widely used for evaluating pharmacokinetic characteristics such as intestinal membrane permeability. However, measurements of the apparent permeability coefficient of highly lipophilic compounds are often limited by transport across the membrane filters, not by transport across the cultured cells. To overcome this concern, we have investigated the utility of a high-porosity membrane honeycomb film (HCF) for transcellular transport studies. Using the HCF inserts, the apparent permeability coefficient (Papp) of the drugs tested in LLC-PK1 and Caco-2 cells tended to increase with an increase in lipophilicity, reaching a maximum Papp value at Log D higher than 2. In contrast, using the commercially available Track-Etched membrane (TEM) inserts, a maximum value was observed at Log D higher than 1. The basolateral to apical transport permeability Papp(BL→AP) of rhodamine 123 across LLC-PK1 cells that express P-glycoprotein (P-gp) cultured on HCF inserts and TEM inserts was 2.33 and 2.39 times higher than the reverse directional Papp(AP→BL) permeability, respectively. The efflux ratio (Papp(B-A)/Papp(A-B)) of rhodamine 123 in LLC-PK1 expressing P-gp cells using HCF inserts was comparable to that obtained using TEM inserts, whereas the transported amount in both directions was significantly higher when using the HCF inserts. Accordingly, due to the higher permeability and high porosity of HCF membranes, it is expected that transcellular transport of high lipophilic as well as hydrophilic compounds and substrate recognition of transporters can be evaluated more accurately by using HCF inserts.

Effect of α-Lipoic Acid on the Development of Human Skin Equivalents Using a Pumpless Skin-on-a-Chip Model.

Owing to the prohibition of cosmetic animal testing, various attempts have recently been made using skin-on-a-chip (SOC) technology as a replacement for animal testing. Previously, we reported the development of a pumpless SOC capable of drug testing with a simple drive using the principle that the medium flows along the channel by gravity when the chip is tilted using a microfluidic channel. In this study, using pumpless SOC, instead of drug testing at the single-cell level, we evaluated the efficacy of α-lipoic acid (ALA), which is known as an anti-aging substance in skin equivalents, for skin tissue and epidermal structure formation. The expression of proteins and changes in genotyping were compared and evaluated. Hematoxylin and eosin staining for histological analysis showed a difference in the activity of fibroblasts in the dermis layer with respect to the presence or absence of ALA. We observed that the epidermis layer became increasingly prominent as the culture period was extended by treatment with 10 µM ALA. The expression of epidermal structural proteins of filaggrin, involucrin, keratin 10, and collagen IV increased because of the effect of ALA. Changes in the epidermis layer were noticeable after the ALA treatment. As a result of aging, damage to the skin-barrier function and structural integrity is reduced, indicating that ALA has an anti-aging effect. We performed a gene analysis of filaggrin, involucrin, keratin 10, integrin, and collagen I genes in ALA-treated human skin equivalents, which indicated an increase in filaggrin gene expression after ALA treatment. These results indicate that pumpless SOC can be used as an in vitro skin model similar to human skin, protein and gene expression can be analyzed, and it can be used for functional drug tests of cosmetic materials in the future. This technology is expected to contribute to the development of skin disease models.

Human iPSC-Based Modeling of Central Nerve System Disorders for Drug Discovery.

A high-throughput drug screen identifies potentially promising therapeutics for clinical trials. However, limitations that persist in current disease modeling with limited physiological relevancy of human patients skew drug responses, hamper translation of clinical efficacy, and contribute to high clinical attritions. The emergence of induced pluripotent stem cell (iPSC) technology revolutionizes the paradigm of drug discovery. In particular, iPSC-based three-dimensional (3D) tissue engineering that appears as a promising vehicle of in vitro disease modeling provides more sophisticated tissue architectures and micro-environmental cues than a traditional two-dimensional (2D) culture. Here we discuss 3D based organoids/spheroids that construct the advanced modeling with evolved structural complexity, which propels drug discovery by exhibiting more human specific and diverse pathologies that are not perceived in 2D or animal models. We will then focus on various central nerve system (CNS) disease modeling using human iPSCs, leading to uncovering disease pathogenesis that guides the development of therapeutic strategies. Finally, we will address new opportunities of iPSC-assisted drug discovery with multi-disciplinary approaches from bioengineering to Omics technology. Despite technological challenges, iPSC-derived cytoarchitectures through interactions of diverse cell types mimic patients' CNS and serve as a platform for therapeutic development and personalized precision medicine.

In Vitro Contraction of Isolated Cauda Epididymal Duct Smooth Muscle as a Complimentary Approach to Physiological, Pathological, Toxicological, and Pharmacological Studies on Epididymal Function.

Contraction of cauda epididymal duct (CE) smooth muscle is one of the very first events of the seminal emission phase of ejaculation. The contraction of CE smooth muscle is governed by a complex interaction of hormones, autacoids, and by the neurotransmitters released from the epididymal intramural nerve endings, and any impairment in the CE smooth muscle contraction has the potential to impair male fertility. Apart the obvious pathophysiological and toxicological importance of CE smooth muscle contraction, modulation of CE contraction has pharmaceutical interest offering a druggable target to development of drugs to improve/impair male fertility. The in vitro contraction experiments constitute a valuable approach to an in-depth evaluation of functional and molecular changes resulting from pathologies or drug exposure. Therefore, this chapter consists in a description of in vitro pharmacological reactivity contractility of the epididymal duct in a controlled medium, maintained at 30 °C of temperature and continuously bubbled with 95% O2 and 5% CO2 to obtain cumulative concentration-response curves that has been fundamental to some of our investigations on epididymal physiology, toxicology, and pharmacology.

A Multicompartment Human Kidney Proximal Tubule-on-a-Chip Replicates Cell Polarization-Dependent Cisplatin Toxicity.

Drug-induced kidney injury is a major clinical problem and causes drug attrition in the pharmaceutical industry. To better predict drug-induced kidney injury, kidney in vitro cultures with enhanced physiologic relevance are developed. To mimic the proximal tubule, the main site of adverse drug reactions in the kidney, human-derived renal proximal tubule epithelial cells (HRPTECs) were injected in one of the channels of dual-channel Nortis chips and perfused for 7 days. Tubes of HRPTECs demonstrated expression of tight junction protein 1 (zona occludens-1), lotus lectin, and primary cilia with localization at the apical membrane, indicating an intact proximal tubule brush border. Gene expression of cisplatin efflux transporters multidrug and toxin extrusion transporter (MATE) 1 (SLC47A1) and MATE2-k (SLC47A2) and megalin endocytosis receptor increased 19.9 ± 5.0-, 23.2 ± 8.4-, and 106 ± 33-fold, respectively, in chip cultures compared with 2-dimensional cultures. Moreover, organic cation transporter 2 (OCT2) (SLC22A2) was localized exclusively on the basolateral membrane. When infused from the basolateral compartment, cisplatin (25 µM, 72 hours) induced toxicity, which was evident as reduced cell number and reduced barrier integrity compared with vehicle-treated chip cultures. Coexposure with the OCT2 inhibitor cimetidine (1 mM) abolished cisplatin toxicity. In contrast, infusion of cisplatin from the apical compartment did not induce toxicity, which was in line with polarized localization of cisplatin uptake transport proteins, including OCT2. In conclusion, we developed a dual channel human kidney proximal tubule-on-a-chip with a polarized epithelium, restricting cisplatin sensitivity to the basolateral membrane and suggesting improved physiologic relevance over single-compartment models. Its implementation in drug discovery holds promise to improve future in vitro drug-induced kidney injury studies. SIGNIFICANCE STATEMENT: Human-derived kidney proximal tubule cells retained characteristics of epithelial polarization in vitro when cultured in the kidney-on-a-chip, and the dual-channel construction allowed for drug exposure using the physiologically relevant compartment. Therefore, cell polarization-dependent cisplatin toxicity could be replicated for the first time in a kidney proximal tubule-on-a-chip. The use of this physiologically relevant model in drug discovery has potential to aid identification of safe novel drugs and contribute to reducing attrition rates due to drug-induced kidney injury.

A combined high-throughput and high-content platform for unified on-chip synthesis, characterization and biological screening.

Acceleration and unification of drug discovery is important to reduce the effort and cost of new drug development. Diverse chemical and biological conditions, specialized infrastructure and incompatibility between existing analytical methods with high-throughput, nanoliter scale chemistry make the whole drug discovery process lengthy and expensive. Here, we demonstrate a chemBIOS platform combining on-chip chemical synthesis, characterization and biological screening. We developed a dendrimer-based surface patterning that enables the generation of high-density nanodroplet arrays for both organic and aqueous liquids. Each droplet (among > 50,000 droplets per plate) functions as an individual, spatially separated nanovessel, that can be used for solution-based synthesis or analytical assays. An additional indium-tin oxide coating enables ultra-fast on-chip detection down to the attomole per droplet by matrix-assisted laser desorption/ionization mass spectrometry. The excellent optical properties of the chemBIOS platform allow for on-chip characterization and in-situ reaction monitoring in the ultraviolet, visible (on-chip UV-Vis spectroscopy and optical microscopy) and infrared (on-chip IR spectroscopy) regions. The platform is compatible with various cell-biological screenings, which opens new avenues in the fields of high-throughput synthesis and drug discovery.

Machine learning classification can reduce false positives in structure-based virtual screening.

With the recent explosion in the size of libraries available for screening, virtual screening is positioned to assume a more prominent role in early drug discovery's search for active chemical matter. In typical virtual screens, however, only about 12% of the top-scoring compounds actually show activity when tested in biochemical assays. We argue that most scoring functions used for this task have been developed with insufficient thoughtfulness into the datasets on which they are trained and tested, leading to overly simplistic models and/or overtraining. These problems are compounded in the literature because studies reporting new scoring methods have not validated their models prospectively within the same study. Here, we report a strategy for building a training dataset (D-COID) that aims to generate highly compelling decoy complexes that are individually matched to available active complexes. Using this dataset, we train a general-purpose classifier for virtual screening (vScreenML) that is built on the XGBoost framework. In retrospective benchmarks, our classifier shows outstanding performance relative to other scoring functions. In a prospective context, nearly all candidate inhibitors from a screen against acetylcholinesterase show detectable activity; beyond this, 10 of 23 compounds have IC50 better than 50 µM. Without any medicinal chemistry optimization, the most potent hit has IC50 280 nM, corresponding to Ki of 173 nM. These results support using the D-COID strategy for training classifiers in other computational biology tasks, and for vScreenML in virtual screening campaigns against other protein targets. Both D-COID and vScreenML are freely distributed to facilitate such efforts.