The Lineage-Defining Transcription Factors SOX2 and NKX2-1 Determine Lung Cancer Cell Fate and Shape the Tumor Immune Microenvironment.

The major types of non-small-cell lung cancer (NSCLC)-squamous cell carcinoma and adenocarcinoma-have distinct immune microenvironments. We developed a genetic model of squamous NSCLC on the basis of overexpression of the transcription factor Sox2, which specifies lung basal cell fate, and loss of the tumor suppressor Lkb1 (SL mice). SL tumors recapitulated gene-expression and immune-infiltrate features of human squamous NSCLC; such features included enrichment of tumor-associated neutrophils (TANs) and decreased expression of NKX2-1, a transcriptional regulator that specifies alveolar cell fate. In Kras-driven adenocarcinomas, mis-expression of Sox2 or loss of Nkx2-1 led to TAN recruitment. TAN recruitment involved SOX2-mediated production of the chemokine CXCL5. Deletion of Nkx2-1 in SL mice (SNL) revealed that NKX2-1 suppresses SOX2-driven squamous tumorigenesis by repressing adeno-to-squamous transdifferentiation. Depletion of TANs in SNL mice reduced squamous tumors, suggesting that TANs foster squamous cell fate. Thus, lineage-defining transcription factors determine the tumor immune microenvironment, which in turn might impact the nature of the tumor.

Presentation of Human Neural Stem Cell Antigens Drives Regulatory T Cell Induction.

Transplantation of human neural stem cells (hNSCs) is a promising regenerative therapy to promote remyelination in patients with multiple sclerosis (MS). Transplantation of hNSCs has been shown to increase the number of CD4+CD25+Foxp3+ T regulatory cells (Tregs) in the spinal cords of murine models of MS, which is correlated with a strong localized remyelination response. However, the mechanisms by which hNSC transplantation leads to an increase in Tregs in the CNS remains unclear. We report that hNSCs drive the conversion of T conventional (Tconv) cells into Tregs in vitro. Conversion of Tconv cells is Ag driven and fails to occur in the absence of TCR stimulation by cognate antigenic self-peptides. Furthermore, CNS Ags are sufficient to drive this conversion in the absence of hNSCs in vitro and in vivo. Importantly, only Ags presented in the thymus during T cell selection drive this Treg response. In this study, we investigate the mechanisms by which hNSC Ags drive the conversion of Tconv cells into Tregs and may provide key insight needed for the development of MS therapies.

Cystatin F attenuates neuroinflammation and demyelination following murine coronavirus infection of the central nervous system.

BACKGROUND: Cystatin F is a secreted lysosomal cysteine protease inhibitor that has been implicated in affecting the severity of demyelination and enhancing remyelination in pre-clinical models of immune-mediated demyelination. How cystatin F impacts neurologic disease severity following viral infection of the central nervous system (CNS) has not been well characterized and was the focus of this study. We used cystatin F null-mutant mice (Cst7-/-) with a well-established model of murine coronavirus-induced neurologic disease to evaluate the contributions of cystatin F in host defense, demyelination and remyelination. METHODS: Wildtype controls and Cst7-/- mice were intracranially (i.c.) infected with a sublethal dose of the neurotropic JHM strain of mouse hepatitis virus (JHMV), with disease progression and survival monitored daily. Viral plaque assays and qPCR were used to assess viral levels in CNS. Immune cell infiltration into the CNS and immune cell activation were determined by flow cytometry and 10X genomics chromium 3' single cell RNA sequencing (scRNA-seq). Spinal cord demyelination was determined by luxol fast blue (LFB) and Hematoxylin/Eosin (H&E) staining and axonal damage assessed by immunohistochemical staining for SMI-32. Remyelination was evaluated by electron microscopy (EM) and calculation of g-ratios. RESULTS: JHMV-infected Cst7-/- mice were able to control viral replication within the CNS, indicating that cystatin F is not essential for an effective Th1 anti-viral immune response. Infiltration of T cells into the spinal cords of JHMV-infected Cst7-/- mice was increased compared to infected controls, and this correlated with increased axonal damage and demyelination associated with impaired remyelination. Single-cell RNA-seq of CD45 + cells enriched from spinal cords of infected Cst7-/- and control mice revealed enhanced expression of transcripts encoding T cell chemoattractants, Cxcl9 and Cxcl10, combined with elevated expression of interferon-g (Ifng) and perforin (Prf1) transcripts in CD8 + T cells from Cst7-/- mice compared to controls. CONCLUSIONS: Cystatin F is not required for immune-mediated control of JHMV replication within the CNS. However, JHMV-infected Cst7-/- mice exhibited more severe clinical disease associated with increased demyelination and impaired remyelination. The increase in disease severity was associated with elevated expression of T cell chemoattractant chemokines, concurrent with increased neuroinflammation. These findings support the idea that cystatin F influences expression of proinflammatory gene expression impacting neuroinflammation, T cell activation and/or glia cell responses ultimately impacting neuroinflammation and neurologic disease.

In Vitro Characterization and Rescue of VX Metabolism in Human Liver Microsomes.

Venomous agent X (VX) is an organophosphate acetylcholinesterase (AChE) inhibitor, and although it is one of the most toxic AChE inhibitors known, the extent of metabolism in humans is not currently well understood. The known metabolism in humans is limited to the metabolite identification from a single victim of the Osaka poisoning in 1994, which allowed for the identification of several metabolic products. VX has been reported to be metabolized in vitro by paraoxonase-1 and phosphotriesterase, although their binding constants are many orders of magnitude above the LD50, suggesting limited physiologic relevance. Using incubation with human liver microsomes (HLMs), we have now characterized the metabolism of VX and the formation of multiple metabolites as well as identified a Food and Drug Administration-approved drug [ethylenediaminetetraacetic acid (EDTA)] that enhances the metabolic rate. HLM incubation alone shows a pronounced increase in the metabolism of VX compared with buffer, suggesting that cytochrome P450-mediated metabolism of VX is occurring. We identified a biphasic decay with two distinct rates of metabolism. The enhancement of VX metabolism in multiple buffers was assessed to attempt to mitigate the effect of hydrolysis rates. The formation of VX metabolites was shown to be shifted with HLMs, suggesting a pathway enhancement over simple hydrolysis. Additionally, our investigation of hydrolysis rates in various common buffers used in biologic assays discovered dramatic differences in VX stability. The new human in vitro VX metabolic data reported points to a potential in vivo treatment strategy (EDTA) for rescue in individuals that are poisoned though enhancement of metabolism alongside existing treatments. SIGNIFICANCE STATEMENT: Venomous agent X (VX) is a potent acetylcholinesterase inhibitor and chemical weapon. To date, we do not possess a clear understanding of its metabolism in humans that would assist us in treating those exposed to it. This study now describes the human liver microsomal metabolism of VX and identifies ethylenediaminetetraacetic acid, which appears to enhance the rate of metabolism. This may provide a potential treatment option for human VX poisoning.

Sequential Contrastive and Deep Learning Models to Identify Selective Butyrylcholinesterase Inhibitors.

Butyrylcholinesterase (BChE) is a target of interest in late-stage Alzheimer's Disease (AD) where selective BChE inhibitors (BIs) may offer symptomatic treatment without the harsh side effects of acetylcholinesterase (AChE) inhibitors. In this study, we explore multiple machine learning strategies to identify BIs in silico, optimizing for precision over all other metrics. We compare state-of-the-art supervised contrastive learning (CL) with deep learning (DL) and Random Forest (RF) machine learning, across single and sequential modeling configurations, to identify the best models for BChE selectivity. We used these models to virtually screen a vendor library of 5 million compounds for BIs and tested 20 of these compounds in vitro. Seven of the 20 compounds displayed selectivity for BChE over AChE, reflecting a hit rate of 35% for our model predictions, suggesting a highly efficient strategy for modeling selective inhibition.

Near-Term Quantum Classification Algorithms Applied to Antimalarial Drug Discovery.

Computational approaches are widely applied in drug discovery to explore properties related to bioactivity, physiochemistry, and toxicology. Over at least the last 20 years, the exploitation of machine learning on molecular data sets has been used to understand the structure-activity relationships that exist between biomolecules and druggable targets. More recently, these methods have also seen application for phenotypic screening data for neglected diseases such as tuberculosis and malaria. Herein, we apply machine learning to build quantum Quantitative Structure Activity Relationship models from antimalarial data sets. There is a continual need for new antimalarials to address drug resistance, and the readily available in vitro data sets could be utilized with newer machine learning approaches as these develop. Furthermore, quantum machine learning is a relatively new method that uses a quantum computer to perform the calculations. First, we present a classical-quantum hybrid computational approach by building a Latent Bernoulli Autoencoder machine learning model for compressing bit-vector descriptors to a size that can be adapted to quantum computers for classification tasks with limited loss of embedded information. Second, we apply our method for feature map compression to quantum classification algorithms, including a completely novel machine learning algorithm with no analogy in classical computers: the Quantum Fourier Transform Classifier. We apply both these approaches to build quantum machine learning models for small-molecule antimalarials with quantum simulation software and then benchmark these quantum models against classical machine learning approaches. While there are many challenges currently facing the development of reliable quantum computers, our results demonstrate that there is potential for the use of this technology in the field of drug discovery.

Enhancement and maximum in the isobaric specific-heat capacity measurements of deeply supercooled water using ultrafast calorimetry.

Knowledge of the temperature dependence of the isobaric specific heat (Cp) upon deep supercooling can give insights regarding the anomalous properties of water. If a maximum in Cp exists at a specific temperature, as in the isothermal compressibility, it would further validate the liquid-liquid critical point model that can explain the anomalous increase in thermodynamic response functions. The challenge is that the relevant temperature range falls in the region where ice crystallization becomes rapid, which has previously excluded experiments. Here, we have utilized a methodology of ultrafast calorimetry by determining the temperature jump from femtosecond X-ray pulses after heating with an infrared laser pulse and with a sufficiently long time delay between the pulses to allow measurements at constant pressure. Evaporative cooling of ∼15-µm diameter droplets in vacuum enabled us to reach a temperature down to ∼228 K with a small fraction of the droplets remaining unfrozen. We observed a sharp increase in Cp, from 88 J/mol/K at 244 K to about 218 J/mol/K at 229 K where a maximum is seen. The Cp maximum is at a similar temperature as the maxima of the isothermal compressibility and correlation length. From the Cp measurement, we estimated the excess entropy and self-diffusion coefficient of water and these properties decrease rapidly below 235 K.

Early-stage dynamics of chloride ion-pumping rhodopsin revealed by a femtosecond X-ray laser.

Chloride ion-pumping rhodopsin (ClR) in some marine bacteria utilizes light energy to actively transport Cl- into cells. How the ClR initiates the transport is elusive. Here, we show the dynamics of ion transport observed with time-resolved serial femtosecond (fs) crystallography using the Linac Coherent Light Source. X-ray pulses captured structural changes in ClR upon flash illumination with a 550 nm fs-pumping laser. High-resolution structures for five time points (dark to 100 ps after flashing) reveal complex and coordinated dynamics comprising retinal isomerization, water molecule rearrangement, and conformational changes of various residues. Combining data from time-resolved spectroscopy experiments and molecular dynamics simulations, this study reveals that the chloride ion close to the Schiff base undergoes a dissociation-diffusion process upon light-triggered retinal isomerization.

COVID-19 and the impact on Alzheimer's disease pathology.

Coronavirus disease 2019 (COVID-19) has rapidly escalated into a global pandemic that primarily affects older and immunocompromised individuals due to underlying clinical conditions and suppressed immune responses. Furthermore, COVID-19 patients exhibit a spectrum of neurological symptoms, indicating that COVID-19 can affect the brain in a variety of manners. Many studies, past and recent, suggest a connection between viral infections and an increased risk of neurodegeneration, raising concerns about the neurological effects of COVID-19 and the possibility that it may contribute to Alzheimer's disease (AD) onset or worsen already existing AD pathology through inflammatory processes given that both COVID-19 and AD share pathological features and risk factors. This leads us to question whether COVID-19 is a risk factor for AD and how these two conditions might influence each other. Considering the extensive reach of the COVID-19 pandemic and the devastating impact of the ongoing AD pandemic, their combined effects could have significant public health consequences worldwide.

Microglia Do Not Restrict SARS-CoV-2 Replication following Infection of the Central Nervous System of K18-Human ACE2 Transgenic Mice.

Unlike SARS-CoV-1 and MERS-CoV, infection with SARS-CoV-2, the viral pathogen responsible for COVID-19, is often associated with neurologic symptoms that range from mild to severe, yet increasing evidence argues the virus does not exhibit extensive neuroinvasive properties. We demonstrate SARS-CoV-2 can infect and replicate in human iPSC-derived neurons and that infection shows limited antiviral and inflammatory responses but increased activation of EIF2 signaling following infection as determined by RNA sequencing. Intranasal infection of K18 human ACE2 transgenic mice (K18-hACE2) with SARS-CoV-2 resulted in lung pathology associated with viral replication and immune cell infiltration. In addition, ∼50% of infected mice exhibited CNS infection characterized by wide-spread viral replication in neurons accompanied by increased expression of chemokine (Cxcl9, Cxcl10, Ccl2, Ccl5 and Ccl19) and cytokine (Ifn-λ and Tnf-α) transcripts associated with microgliosis and a neuroinflammatory response consisting primarily of monocytes/macrophages. Microglia depletion via administration of colony-stimulating factor 1 receptor inhibitor, PLX5622, in SARS-CoV-2 infected mice did not affect survival or viral replication but did result in dampened expression of proinflammatory cytokine/chemokine transcripts and a reduction in monocyte/macrophage infiltration. These results argue that microglia are dispensable in terms of controlling SARS-CoV-2 replication in in the K18-hACE2 model but do contribute to an inflammatory response through expression of pro-inflammatory genes. Collectively, these findings contribute to previous work demonstrating the ability of SARS-CoV-2 to infect neurons as well as emphasizing the potential use of the K18-hACE2 model to study immunological and neuropathological aspects related to SARS-CoV-2-induced neurologic disease. IMPORTANCE Understanding the immunological mechanisms contributing to both host defense and disease following viral infection of the CNS is of critical importance given the increasing number of viruses that are capable of infecting and replicating within the nervous system. With this in mind, the present study was undertaken to evaluate the role of microglia in aiding in host defense following experimental infection of the central nervous system (CNS) of K18-hACE2 with SARS-CoV-2, the causative agent of COVID-19. Neurologic symptoms that range in severity are common in COVID-19 patients and understanding immune responses that contribute to restricting neurologic disease can provide important insight into better understanding consequences associated with SARS-CoV-2 infection of the CNS.

Human CYP2C19 Substrate and Inhibitor Characterization of Organophosphate Pesticides.

CYP2C19 is an important enzyme for organophosphate pesticide (OPP) metabolism. Because the OPPs can be both substrates and inhibitors of CYP2C19, we screened 45 OPPs for their ability to inhibit the activity of this enzyme and investigated the role of CYP2C19 in the metabolism of 22 of these molecules. We identified several nanomolar inhibitors of CYP2C19 as well as determined that thions, in general, are more potent inhibitors than oxons. We also determined that thions are readily metabolized by CYP2C19, although we saw no relationship between IC50 values and intrinsic clearance rates. This study may have implications for mitigating the risk of OPP poisoning.

Validation of Acetylcholinesterase Inhibition Machine Learning Models for Multiple Species.

Acetylcholinesterase (AChE) is an important enzyme and target for human therapeutics, environmental safety, and global food supply. Inhibitors of this enzyme are also used for pest elimination and can be misused for suicide or chemical warfare. Adverse effects of AChE pesticides on nontarget organisms, such as fish, amphibians, and humans, have also occurred as a result of biomagnifications of these toxic compounds. We have exhaustively curated the public data for AChE inhibition data and developed machine learning classification models for seven different species. Each set of models were built using up to nine different algorithms for each species and Morgan fingerprints (ECFP6) with an activity cutoff of 1 µM. The human (4075 compounds) and eel (5459 compounds) consensus models predicted AChE inhibition activity using external test sets from literature data with 81% and 82% accuracy, respectively, while the reciprocal cross (76% and 82% percent accuracy) was not species-specific. In addition, we also created machine learning regression models for human and eel AChE inhibition to return a predicted IC50 value for a queried molecule. We did observe an improved species specificity in the regression models, where a human support vector regression model of human AChE inhibition (3652 compounds) predicted the IC50s of the human test set to a better extent than the eel regression model (4930 compounds) on the same test set, based on mean absolute percentage error (MAPE = 9.73% vs 13.4%). The predictive power of these models certainly benefits from increasing the chemical diversity of the training set, as evidenced by expanding our human classification model by incorporating data from the Tox21 library of compounds. Of the 10 compounds we tested that were predicted active by this expanded model, two showed >80% inhibition at 100 µM. This machine learning approach therefore offers the ability to rapidly score massive libraries of molecules against the models for AChE inhibition that can then be selected for future in vitro testing to identify potential toxins. It also enabled us to create a public website, MegaAChE, for single-molecule predictions of AChE inhibition using these models at megaache.collaborationspharma.com.

In silico ADME/tox comes of age: twenty years later.

In the early 2000s pharmaceutical drug discovery was beginning to use computational approaches for absorption, distribution, metabolism, excretion and toxicity (ADME/Tox, also known as ADMET) prediction. This emphasis on prediction was an effort to reduce the risk of later stage failures from ADME/Tox.Much has been written in the intervening twenty plus years and significant expenditure has occurred in companies developing these in silico capabilities which can be gleaned from publications. It is therefore an appropriate time to briefly reflect on what was proposed then and what the reality is today.20 years ago, we tended to optimise bioactivity and perhaps one ADME/Tox property at a time. Previously pharmaceutical companies needed a whole infrastructure for models - in silico and in vitro experts, IT, champions on a project team, educators and management support. Now we are in the age of generative de novo design where bioactivity and many ADME/Tox properties can be optimised and large language model technologies are available.There are also some challenges such as the focus on very large molecules which may be outside of current ADME/Tox models.We provide an opportunity to look forward with the increasing public data for ADME/Tox as well as expanded types of algorithms available.

MAC2 is a long-lasting marker of peripheral cell infiltrates into the mouse CNS after bone marrow transplantation and coronavirus infection.

Microglia are the primary resident myeloid cells of the brain responsible for maintaining homeostasis and protecting the central nervous system (CNS) from damage and infection. Monocytes and monocyte-derived macrophages arising from the periphery have also been implicated in CNS pathologies, however, distinguishing between different myeloid cell populations in the CNS has been difficult. Here, we set out to develop a reliable histological marker that can assess distinct myeloid cell heterogeneity and functional contributions, particularly in the context of disease and/or neuroinflammation. scRNAseq from brains of mice infected with the neurotropic JHM strain of the mouse hepatitis virus (JHMV), a mouse coronavirus, revealed that Lgals3 is highly upregulated in monocyte and macrophage populations, but not in microglia. Subsequent immunostaining for galectin-3 (encoded by Lgals3), also referred to as MAC2, highlighted the high expression levels of MAC2 protein in infiltrating myeloid cells in JHMV-infected and bone marrow (BM) chimeric mice, in stark contrast to microglia, which expressed little to no staining in these models. Expression of MAC2 was found even 6-10 months following BM-derived cell infiltration into the CNS. We also demonstrate that MAC2 is not a specific label for plaque-associated microglia in the 5xFAD mouse model, but only appears in a distinct subset of these cells in the presence of JHMV infection or during aging. Our data suggest that MAC2 can serve as a reliable and long-lasting histological marker for monocyte/macrophages in the brain, identifying an accessible approach to distinguishing resident microglia from infiltrating cells in the CNS under certain conditions.

Acoustic Focusing of Protein Crystals for In-Line Monitoring and Up-Concentration during Serial Crystallography.

Serial femtosecond crystallography (SFX) has become one of the standard techniques at X-ray free-electron lasers (XFELs) to obtain high-resolution structural information from microcrystals of proteins. Nevertheless, reliable sample delivery is still often limiting data collection, as microcrystals can clog both field- and flow-focusing nozzles despite in-line filters. In this study, we developed acoustic 2D focusing of protein microcrystals in capillaries that enables real-time online characterization of crystal size and shape in the sample delivery line after the in-line filter. We used a piezoelectric actuator to create a standing wave perpendicular to the crystal flow, which focused lysozyme microcrystals into a single line inside a silica capillary so that they can be imaged using a high-speed camera. We characterized the acoustic contrast factor, focus size, and the coaxial flow lines and developed a splitting union that enables up-concentration to at least a factor of five. The focus size, flow rates, and geometry may enable an upper limit of up-concentration as high as 200 fold. The novel feedback and concentration control could be implemented for serial crystallography at synchrotrons with minor modifications. It will also aid the development of improved sample delivery systems that will increase SFX data collection rates at XFELs, with potential applications to many proteins that can only be purified and crystallized in small amounts.

EphA2 contributes to disruption of the blood-brain barrier in cerebral malaria.

Disruption of blood-brain barrier (BBB) function is a key feature of cerebral malaria. Increased barrier permeability occurs due to disassembly of tight and adherens junctions between endothelial cells, yet the mechanisms governing junction disassembly and vascular permeability during cerebral malaria remain poorly characterized. We found that EphA2 is a principal receptor tyrosine kinase mediating BBB breakdown during Plasmodium infection. Upregulated on brain microvascular endothelial cells in response to inflammatory cytokines, EphA2 is required for the loss of junction proteins on mouse and human brain microvascular endothelial cells. Furthermore, EphA2 is necessary for CD8+ T cell brain infiltration and subsequent BBB breakdown in a mouse model of cerebral malaria. Blocking EphA2 protects against BBB breakdown highlighting EphA2 as a potential therapeutic target for cerebral malaria.

Rational Control of Off-State Heterogeneity in a Photoswitchable Fluorescent Protein Provides Switching Contrast Enhancement.

Reversibly photoswitchable fluorescent proteins are essential markers for advanced biological imaging, and optimization of their photophysical properties underlies improved performance and novel applications. Here we establish a link between photoswitching contrast, one of the key parameters that dictate the achievable resolution in nanoscopy applications, and chromophore conformation in the non-fluorescent state of rsEGFP2, a widely employed label in REversible Saturable OpticaL Fluorescence Transitions (RESOLFT) microscopy. Upon illumination, the cis chromophore of rsEGFP2 isomerizes to two distinct off-state conformations, trans1 and trans2, located on either side of the V151 side chain. Reducing or enlarging the side chain at this position (V151A and V151L variants) leads to single off-state conformations that exhibit higher and lower switching contrast, respectively, compared to the rsEGFP2 parent. The combination of structural information obtained by serial femtosecond crystallography with high-level quantum chemical calculations and with spectroscopic and photophysical data determined in vitro suggests that the changes in switching contrast arise from blue- and red-shifts of the absorption bands associated to trans1 and trans2, respectively. Thus, due to elimination of trans2, the V151A variants of rsEGFP2 and its superfolding variant rsFolder2 display a more than two-fold higher switching contrast than their respective parent proteins, both in vitro and in E. coli cells. The application of the rsFolder2-V151A variant is demonstrated in RESOLFT nanoscopy. Our study rationalizes the connection between structural and photophysical chromophore properties and suggests a means to rationally improve fluorescent proteins for nanoscopy applications.

Machine Learning Models Identify New Inhibitors for Human OATP1B1.

The uptake transporter OATP1B1 (SLC01B1) is largely localized to the sinusoidal membrane of hepatocytes and is a known victim of unwanted drug-drug interactions. Computational models are useful for identifying potential substrates and/or inhibitors of clinically relevant transporters. Our goal was to generate OATP1B1 in vitro inhibition data for [3H] estrone-3-sulfate (E3S) transport in CHO cells and use it to build machine learning models to facilitate a comparison of seven different classification models (Deep learning, Adaboosted decision trees, Bernoulli naïve bayes, k-nearest neighbors (knn), random forest, support vector classifier (SVC), logistic regression (lreg), and XGBoost (xgb)] using ECFP6 fingerprints to perform 5-fold, nested cross validation. In addition, we compared models using 3D pharmacophores, simple chemical descriptors alone or plus ECFP6, as well as ECFP4 and ECFP8 fingerprints. Several machine learning algorithms (SVC, lreg, xgb, and knn) had excellent nested cross validation statistics, particularly for accuracy, AUC, and specificity. An external test set containing 207 unique compounds not in the training set demonstrated that at every threshold SVC outperformed the other algorithms based on a rank normalized score. A prospective validation test set was chosen using prediction scores from the SVC models with ECFP fingerprints and were tested in vitro with 15 of 19 compounds (84% accuracy) predicted as active (≥20% inhibition) showed inhibition. Of these compounds, six (abamectin, asiaticoside, berbamine, doramectin, mobocertinib, and umbralisib) appear to be novel inhibitors of OATP1B1 not previously reported. These validated machine learning models can now be used to make predictions for drug-drug interactions for human OATP1B1 alongside other machine learning models for important drug transporters in our MegaTrans software.

Machine Learning Models for Mycobacterium tuberculosisIn Vitro Activity: Prediction and Target Visualization.

Tuberculosis (TB) is a major global health challenge, with approximately 1.4 million deaths per year. There is still a need to develop novel treatments for patients infected with Mycobacterium tuberculosis (Mtb). There have been many large-scale phenotypic screens that have led to the identification of thousands of new compounds. Yet, there is very limited investment in TB drug discovery which points to the need for new methods to increase the efficiency of drug discovery against Mtb. We have used machine learning approaches to learn from the public Mtb data, resulting in many data sets and models with robust enrichment and hit rates leading to the discovery of new active compounds. Recently, we have curated predominantly small-molecule Mtb data and developed new machine learning classification models with 18 886 molecules at different activity cutoffs. We now describe the further validation of these Bayesian models using a library of over 1000 molecules synthesized as part of EU-funded New Medicines for TB and More Medicines for TB programs. We highlight molecular features which are enriched in these active compounds. In addition, we provide new regression and classification models that can be used for scoring compound libraries or used to design new molecules. We have also visualized these molecules in the context of known molecular targets and identified clusters in chemical property space, which may aid in future target identification efforts. Finally, we are also making these data sets publicly available, representing a significant increase to the available Mtb inhibition data in the public domain.