Computational Drug Discovery Methods

Q-BioLiP improves upon BioLiP by representing protein-ligand interactions using quaternary structures and providing both experimental and predicted binding affinities.

This paper introduces a computational pipeline for identifying SARS-CoV-2 nsp14 inhibitors using virtual screening, identifying seven candidates including NSC620333 and SS148, with experimental validation of binding affinity and structural insights.

The paper introduces a benchmark dataset and baseline models for few-shot prediction of small molecule activity using cell microscopy images.

This paper introduces a multi-modal deep learning algorithm for drug response prediction that improves generalizability by integrating diverse cell line data and addressing data skewness.

AIxFuse generates dual-target drugs by learning pharmacophore fusion patterns that satisfy structural constraints from molecular docking, outperforming existing methods in success rate and binding free energy.

This study identifies novel SARS-CoV-2 Mpro inhibitors using deep reinforcement learning and computational chemistry methods, showing potent inhibitory activity with IC50 values between 1.3 and 2.3 uM.

This paper reviews drug promiscuity, discussing its role in side effects and therapeutic potential, and proposes a computer modeling strategy to leverage it.

This paper designs synthetic peptide substrates for TAM family kinases using phosphoproteomics and AlphaFold2, showing that predicted interactions correlate with biochemical validation.

ChemEM docks small molecules into medium-resolution cryo-EM maps using difference maps and a scoring function, outperforming existing methods and demonstrating utility at low resolutions.

GeNNius uses a graph neural network to predict drug-target interactions more efficiently and accurately than existing methods across multiple datasets.

This study identifies three phytochemicals from Houttuynia cordata as potential inhibitors of human glutathione reductase through molecular docking and dynamics simulations.

This study analyzes the structure-activity landscape of environmental chemicals binding to multiple human endocrine receptors to improve toxicity prediction models.

This study uses transfer learning with ligand- and structure-based descriptors to improve opioid bioactivity prediction across OR subtypes.

The paper designs and optimizes Y-shaped inhibitors of METTL3 using de novo methods and brute force screening, achieving high selectivity and binding affinity for potential cancer treatment.

This study analyzes drug off-target effects on cervical cancer isoforms, highlighting risks of ignoring isoform-specific interactions in drug design.

PDBe CCDUtils processes and analyzes small molecules from the PDB using RDKit, offering tools for property computation and database cross-references.

This paper introduces MultiPPIMI, a deep learning framework that predicts PPI-modulator interactions without requiring target structure or reference modulators, achieving high AUROC scores in benchmark tests.

This study identifies four potential CDK1 inhibitors through molecular docking and dynamics simulations for colorectal cancer treatment.

CurveCurator assesses and classifies dose-response curve significance using a recalibrated F-statistic and provides p-values and false discovery rates for improved analysis.

The paper identifies a first-in-class small molecule ligand for WDR91 using DNA-encoded chemical library selection and machine learning, leading to the discovery of covalent analogs and structural insights.

This study predicts SPI binding sites on Pol II-Spt5 and shows how strong binding affects RNA and DNA dynamics at the exit site.

The paper evaluates deep learning methods for predicting protein-ligand binding affinity by analyzing the impact of different encodings for proteins and ligands.

The paper links kinase inhibitor treatment effects to cell line responses using integrated kinome profiling data and builds predictive models with high accuracy.

The paper describes an algorithm using deep learning to develop small molecule drugs for lung cancer, identifying target genes and predicting inhibitor interactions and efficacy.

This paper introduces a novel in silico scaffold-hopping method for drug repositioning, identifying 144 compounds with potential new therapeutic applications for rare diseases.

The paper uses machine learning to identify chemical features associated with stage-specific antimalarial activity against Plasmodium falciparum asexual blood stages and gametocytes.

AnimalGAN generates synthetic rat clinical pathology data to assess drug hepatotoxicity, outperforming traditional methods and aligning with human population findings.

This paper identifies potential FDA-approved drugs as inhibitors of Pfmrk, a Plasmodium falciparum protein, using in silico methods.

The paper develops new cyclic compounds targeting the PSD-95 PDZ3 domain through structure-based design, achieving up to 7-fold stronger affinity than a prototype compound.

This paper benchmarks small molecule feature representations for predicting cardiotoxicity against hERG, Nav1.5, and Cav1.2 channels using a new dataset and tool.