Computational Drug Discovery Methods

BaCNet uses deep learning to predict antibiotic-target interactions and discover novel compounds against drug-resistant pathogens.

The paper uses machine learning to predict drug responses in colorectal cancer cells, identifying novel compounds with cytotoxic or cytostatic effects.

MolGene-E generates high-quality drug molecules from single-cell transcriptomics data, outperforming baselines in zero-shot molecule generation.

NRGRank is a fast, accurate virtual screening tool that outperforms docking methods on some targets and complements them on others, enabling efficient large-scale screening.

The paper evaluates the accuracy of in silico, in vitro, and in vivo ADME/PK predictions for drug discovery, finding poor performance and highlighting the superiority of the ANDROMEDA software.

This paper develops a neuro-symbolic AI model using LTN to predict hERG inhibition, outperforming existing models on benchmark datasets.

This study evaluates Unani phytochemicals against HMPV using computational methods, finding strong binding affinities and stability for Crocin, Quercetin, and Oleuropein.

This paper introduces DeepAVC, a framework using pre-trained large language models to predict broad-spectrum antiviral compounds, achieving high accuracy and identifying novel compounds with experimental validation.

This paper introduces PCL analysis to predict small molecule mechanisms of action in Mycobacterium tuberculosis using PROSPECT data, achieving high sensitivity and precision in MOA prediction.

GerNA-Bind predicts RNA-ligand binding specificity using a geometric deep learning framework, achieving improved precision and identifying compounds with experimentally confirmed activity against MALAT1 RNA.

The paper compares MPP+ and dopamine binding to DAT in humans and zebrafish using docking analysis, showing similar affinities and binding residues.

LANTERN improves molecular interaction prediction by integrating large language models and transformers, achieving state-of-the-art results on DTI, DDI, and PPI tasks.

The paper predicts mutation-induced changes in protein-ligand binding affinity using few-shot learning with AlphaFold2 subsampling and Siamese learning, achieving improved correlation on TKI datasets.

The paper introduces a graph representation that combines static and dynamic protein data, improving performance in three tasks related to protein-ligand interactions.

The paper develops a hybrid modeling pipeline to identify antitubercular compounds with favorable potency and pharmacokinetics, resulting in 12 promising candidates from a large compound library.

This study identifies six small molecule compounds that inhibit Ebola virus VP35 and VP40 proteins through virtual screening, molecular dynamics, and ADMET analysis.

This study uses molecular docking on AlphaFold2-predicted protein structures to identify potential targets for puberulic acid, highlighting sodium/myo-inositol cotransporters as high-affinity candidates.

This paper develops a compound generative pre-trained transformer model trained on 200 million compounds to generate chemically valid and novel drug-like molecules with high novelty.

The paper investigates the range of actions of neurosteroid analogues using in silico and in vitro methods, revealing unexpected modulation of multiple targets beyond GABA-A receptors.

This paper introduces ChemBERTaDDI, a framework using transformers to predict drug-drug interactions by combining molecular structures and clinical data, showing improved performance over existing methods.

BoPep identifies peptide binders to specific proteins by efficiently navigating sequence space, reducing docking evaluations and discovering novel therapeutic peptides.

This paper predicts off-target proteins of SGLT2 inhibitors using a bioinformatics approach, identifying structural features that guide binding and drug specificity.

This study uses GALILEO and ChemPrint to generate novel antiviral compounds targeting the conserved Thumb-1 site, achieving 100% hit rate in vitro against HCV and Coronavirus 229E.

This study calculates polarizability tensors for drug-like molecules and uses machine learning to predict IC50s, showing induced polarizability's importance in binding.

This study identifies Remdesivir and Peramivir as promising FDA-approved antiviral drugs for HMPV treatment using computational methods.

This paper presents a computational drug repurposing framework that identifies FDA-approved drugs for COVID-19 by integrating genetic expression data and EMR data, finding four compounds with protective effects in validation.

The paper screens 14.6 million virtual isoquinuclidines against opioid receptors to identify dual antagonists-inverse agonists with reduced withdrawal effects compared to naloxone.

This study compares two transfer learning approaches for predicting small molecule bioactivity, finding similar performance but different strengths in handling smaller assays.

This paper investigates the ADME properties, toxicity, and DNA binding of indole alkaloids from Nagaland plants using computational methods.

This paper introduces CoNCISE, a method that accelerates drug-target interaction prediction by compressing small molecule data into 15 bits per compound, achieving high accuracy and enabling rapid screening of large chemical libraries.