Computational Drug Discovery Methods

This study uses computational methods to analyze how air pollutants interact with the PXR receptor, identifying key compounds and their binding mechanisms.

This study identifies a novel compound with therapeutic potential for female hormonal disorders through computational methods and molecular analysis.

The paper uses machine learning to link molecular structure, bitterness, and mechanism to antimalarial activity, identifying promising chemistries for malaria treatment.

The paper identifies PAINS in a USP8 inhibitor screening campaign for Parkinson's disease and discusses interference mechanisms affecting assay reliability.

This study uses computational methods to evaluate the blood coagulant potential of Caenothus americanus compounds by analyzing docking, physicochemical, and ADMET properties.

This study identifies potential EGFR inhibitors through virtual screening and deep learning models, finding nine compounds that outperform existing drugs in bioactivity and druglikeness.

This paper shows that current deep learning models for predicting kinase inhibitor binding affinities do not generalize well to unseen data, suggesting they rely on data leakage rather than learning molecular interactions.

The paper describes the development of MPDSCOVID-19, a disease-specific web portal for computational drug discovery, with updated modules and tools for COVID-19 research.

SYN-FUSION combines Graph Neural Networks and Transformers to improve molecular property prediction, outperforming existing models on several benchmarks.

The paper computationally designs two novel μ-opioid receptor agonists with improved docking scores and ADMET properties compared to Hydrocodone.

ProSmith predicts drug-target interactions using a multi-modal Transformer Network, showing better generalization and performance than previous models.

This paper introduces GATMDA, a graph attention network with multiple kernel fusion to predict human microbe-drug associations, achieving the highest AUC and AUPR on benchmark datasets.

The paper introduces an E(n) Equivariant Graph Neural Network to predict ligand-protein binding structures, showing promising results for drug design.

The paper identifies FDA-approved drugs that may bind to structurally similar protein-protein interfaces in the Ras/Raf/MEK/ERK pathway, increasing structural coverage from 21% to 92% using PRISM.

This paper introduces E-DIS, a framework for drug-target interaction prediction that improves generalization to out-of-distribution data by combining domain-specific and domain-generic features.

The paper predicts cancer cell line sensitivity to kinase inhibitor combinations using kinome inhibition states and machine learning, achieving high accuracy in both simulated and experimental settings.

This study identifies potential antiviral drugs against influenza virus by computationally prioritizing drug-target pairs and validating their activity in cell lines.

The paper curates a large RNA-targeted small molecule library and identifies novel RNA-binding compounds through in vitro assays.

The paper introduces a causal inference and graph neural network framework for drug repurposing, demonstrating its effectiveness through simulations and application to L1000 connectivity map data.

The paper develops two novel active learning batch selection methods for drug discovery, showing significant improvements over existing methods across multiple datasets.

This study develops dynamic structure-based pharmacophore models for virtual screening of small molecule libraries targeting YB-1, improving upon static models by capturing binding dynamics.

This paper investigates the potential of epicatechin gallate to target aberrant cardiac Ca2+ signaling proteins through in silico analyses and molecular docking.

This study evaluates using CNN-extracted latent representations from toxicopathological images to estimate compound properties in vivo, finding that layer selection significantly impacts performance.

The paper screens natural compounds to identify Asiatic Acid as a potential inhibitor of the R248Q mutant p53, showing significant binding and interactions through molecular dynamics simulations.

This paper reproduces and improves the reusability of the TCRP method for predicting drug response across preclinical and clinical datasets.

This paper introduces SynerGPT, a model that uses in-context learning to predict drug synergy and design personalized cancer treatments based on patient biopsy data.

This study evaluates the interaction between a tetrahydrocurcumin derivative and PCSK9 protein using computational methods, showing improved binding affinity compared to atorvastatin.

This study identifies Quercetin-3-(6-Malonylglucoside) as a potent Bcl-2 inhibitor from Morus alba through molecular docking and simulations, suggesting its potential for cancer chemotherapy.

A GPT-2 retrained from scratch on protein-ligand binding data, with a BPE tokenizer redesigned for proteins and ligands, is proposed for generating ligands against a target protein.