Search

Nuclear Magnetic Resonance (NMR) is crucial for structure determination, yet automated workflows remain underdeveloped compared to mass spectrometry (MS). Here, the authors introduce NMR molecular networking for HSQC spectra, adapting core principle of MS² networking, enhancing annotation and dereplication through innovative algorithms, which significantly improve the identification of unknown metabolites.

Enterololin is a narrow-spectrum antibiotic that selectively kills Enterobacteriaceae in vitro and in a mouse model of adherent-invasive Escherichia coli gut infection. AI-guided mechanism-of-action studies identified LolCDE as its molecular target.

This study presents CSNN, a tool leveraging network homophily and training-free graph neural networks with labels as features to predict drug-target-interactions (DTIs). The model is then experimentally validated on a new dataset of 3773 DTIs from a yeast-based screen on 7 human GPCRs.

An explainable deep learning model using a chemical substructure-based approach for the exploration of chemical compound libraries identified structural classes of compounds with antibiotic activity and low toxicity.

The precise relationship between neurobehavioural effects and neurotransmitter effects of psychiatric drugs are not always understood. Here the authors develop a database documenting the neurotransmitter response in rats to 258 different neuropsychiatric drugs, and conclude that this neurotransmitter response does not, in general, reflect the current categorisation of those drugs.

The IDG-DREAM Challenge carried out crowdsourced benchmarking of predictive algorithms for kinase inhibitor activities on unpublished data. This study provides a resource to compare emerging algorithms and prioritize new kinase activities to accelerate drug discovery and repurposing efforts.

Using a neural network trained on bacterial growth inhibition data, in silico prediction of molecules with activity against Acinetobacter baumannii led to the identification of the narrow-spectrum abaucin, which perturbs lipoprotein trafficking.

Docking virtual libraries against protein structures has identified potent ligands for multiple targets. A comprehensive analysis reveals that the increased size of virtual libraries improves receptor fit but diverges from bio-like molecules.

Ligand detection is an essential component of pre-clinical drug development. Here, the authors develop an assay directly coupling protein ligand binding to the luminescence intensity of NanoLuc luciferase; enabling a compound discovery strategy for a broad spectrum of potential drug targets.

Here, the authors introduce a computational strategy for antimicrobial discovery that addresses the scarcity of large datasets. Based on data-driven representations of molecular structures, this framework uncovers new growth inhibitors of Staphylococcus aureus.

It is unclear whether new antimalarial compounds may rapidly lose effectiveness in the field because of parasite resistance. Here, Corey et al.investigate the acquisition of drug resistance and the extent to which common resistance mechanisms decrease susceptibility to a diverse set of 50 antimalarial compounds.

A computational screen of an ultra-large virtual library against the structure of the melatonin receptor found nanomolar ligands, and ultimately two selective MT₁ inverse agonists that induced phase advancement of the mouse circadian clock when given at subjective dusk.

V-SYNTHES, a scalable and computationally cost-effective synthon-based approach to compound screening, identified compounds with a high affinity for CB2 and CB1 in a hierarchical structure-based screen of more than 11 billion compounds.

Bond et al. predict mechanism of action of hit compounds from a pooled screen of Mycobacterium tuberculosis mutants underproducing essential proteins by comparing the strain-specific responses of screening hits to those elicited by known antimicrobials.

Tummino et al. dock 74 million molecules against the human cannabinoid-1 receptor to find uM ligands. Optimization led to a nM agonist conferring analgesia with reduced side effects in mice, highlighting its potential as a pain therapeutic and the promise of a structure-based approach.

Direct air capture (DAC) of carbon dioxide is a critical technology for climate change mitigation, yet current materials lack efficiency and scalability. Here, the authors employ machine learning and high-throughput atomistic modeling to identify binding sites for the targeted design of functionalized amine sorbents, advancing the design of scalable, cost-effective DAC technologies.

Analgesia from the expectation of pain relief is mediated by rostral anterior cingulate cortex neurons that project to the pontine nucleus.

Non-addictive treatments for pain are much needed. Here, the authors identify in vivo active leads for inflammatory pain using large library docking against the EP4 prostaglandin receptor.

The use of allosteric transcription factors (aTFs) as biosensors has been constrained by their limited natural ligand repertoire. Here, the authors report a method to screen large libraries of aTF variants to develop biosensors with altered specificities to non-native ligands.

A modification of the in silico screening tool, DOCK, allows for identification of compounds that covalently modify catalytic and noncatalytic protein nucleophiles to modulate the activities of bacterial β-lactamase and the kinases RSK2, MSK1 and JAK3.

The docking of a 1.7 billion- versus a 99 million-molecule virtual library against β-lactamase revealed that the larger-sized library produced improved hit rates and potency along with an increased number of scaffolds.

Some anesthetics despite being generally associated with sedation, can also increase brain activity—a phenomenon called paradoxical excitation. The authors identified dozens of compounds that generally decrease neuronal activity, but increase activity in the caudal hindbrain of zebrafish.

Here, a library of more than 300,000 chemicals was screened for activity against Plasmodium falciparum growing in red blood cells. Of these chemicals, 172 representative candidates were profiled in detail; one exemplar compound showed efficacy in a mouse model of malaria. The findings provide the scientific community with new starting points for drug discovery.

Systematic phenotyping of 58,101 mutants of the model eukaryotic alga Chlamydomonas reinhardtii under 121 environmental and chemical stress conditions provides a large resource for characterizing gene function.

Using a make-on-demand library that contains hundreds-of-millions of molecules, structure-based docking was used to identify compounds that, after synthesis and testing, are shown to interact with AmpC β-lactamase and the D₄ dopamine receptor with high affinity.

The adoption of multiple conformations by proteins presents a challenge for ligand discovery using docking simulations. Now, a method for representing the conformational behaviour of a flexible protein in docking screens, which is guided by experimental crystallography data, is shown to predict protein conformation, ligand pose and aid the discovery of new ligands.

Protein/protein interaction (PPI) interfaces have emerged as drug targets to develop medications with less side effects. Here, Dvorak et al. show that small molecule targeting of PPIs in the brain is an approach for psychiatric drug discovery.

A virtual screen of the GPCR D3R based on a homology model prior to publication of the crystal structure and a subsequent virtual screen based on the crystal structure of the receptor once it became available both identified new ligands with verified activities.

Identifying molecular properties of compounds that best correlate with outer membrane permeation and growth inhibition could guide the discovery of new antibiotics. Here, the authors evaluate 174 molecular descriptors in 1260 antimicrobial compounds and study their correlations with antibacterial activity in Gram-negative Pseudomonas aeruginosa to derive a statistical protocol to identify mechanistic predictors of outer membrane permeation.

Computational docking to the the μ-opioid-receptor identifies PZM21, a novel selective biased agonist that generates substantial affective analgesia in mice without altering respiration or inducing drug reinforcement.

Adhirons are small antibody-like proteins which have been isolated against over 350 targets including Crimean-Congo Haemorrhagic Fever Virus and Influenza A Virus. These adhirons can be templates to discover small molecules modulating these targets.

A combination of clinical and molecular analyses supports an association between disorders of immune complement or coagulation with poor outcome in patients with SARS-CoV-2 infection.

A high-throughput chemical–genetic screening approach for the discovery of targets and chemicals to treat Mycobacterium tuberculosis yields tenfold more hit compounds than conventional whole-cell screening methods.

Structural studies of the itch receptors MRGPRX2 and MRGPRX4 in complex with endogenous and synthetic ligands provide a basis for the development of therapeutic compounds for pain, itch and mast cell-mediated hypersensitivity.

New agonists of the serotonin 5-HT_2A receptor (5-HT_2AR) confer high brain permeability and antidepressant activity and—in contrast to classic 5-HT_2AR agonists—lack psychedelic activity.

The mechanical properties of heterogeneous cell populations in colorectal tumors and the relevance to cancer metastasis remain not fully understood. Here, the authors suggest that the variations in malignant phenotypes between LGR5-positive cancer stem cells and LGR5-negative cells could be due to their distinct mechanical phenotypes observed in vitro, determined by the membrane to cortex attachment proteins Ezrin/Radixin/Moesin.