Computational Drug Discovery Methods

The paper uses quantitative systems pharmacology to analyze 50 drugs of abuse, identifying new targets and pathways, with mTORC1 highlighted as a key effector in neuronal restructuring.

DeepAffinity predicts compound-protein affinity using a semi-supervised deep learning model combining recurrent and convolutional networks, achieving high accuracy and interpretability.

The paper identifies potential DNA repair protein inhibitors using computational docking of small molecules associated with cancer survival, highlighting promising drug-like ligands for further experimental validation.

The paper uses tensor factorization to predict successful therapeutic target-indication pairs by analyzing clinical trial data and comparing performance with other machine learning models.

The paper develops a workflow to predict protease-specific cleavage sites using physicochemical properties of amino acids, trained on human proteases and compared to existing tools.

The paper evaluates the CANDO platform for drug repurposing by identifying optimal protein subsets that improve accuracy while reducing computational complexity.

CDEK is a database of active pharmaceutical ingredients with clinical testing evidence and associated organizations, curated from ClinicalTrials.gov and other drug databases.

This study identifies 20 approved drugs that could inhibit Schistosoma mansoni energy metabolism enzymes through chemogenomic analysis of drug-target interactions.

The paper develops a pipeline to predict protein targets for drug-like compounds using transcriptomics and machine learning, achieving 26% top-10 accuracy on FDA-approved drugs.

This paper develops a machine learning method to predict drug toxicity by integrating drug targets, functional impact scores, and biological networks, showing it can identify toxic drugs and support safe combination design.

This paper develops a computational strategy to identify novel targets and therapeutic compounds for opioid dependence using time course gene expression data from mouse striatum.

This study develops a virtual drug screening tool to discover new CDK8 type II ligands, identifying a potent compound with a Kd value of 42.5 nM.

The paper develops NN-MT, a multi-task SVM algorithm for predicting drug-protein interactions, showing efficient performance in proteome-wide and orphan case predictions.

This study benchmarks homology models of human GPCRs for virtual ligand screening, finding that 10 out of 19 GPCRs show comparable performance to crystal structures, with consensus methods improving results.

NeoDTI predicts drug-target interactions by integrating heterogeneous network data, showing improved performance and novel predictions supported by prior studies.

The paper introduces a VR software pipeline for visualizing protein-drug interactions and simulating molecular motions to improve drug design.

This paper introduces a recommender system approach to predict cancer drug responses using pharmacogenomic data and demonstrates its effectiveness across multiple datasets.

The paper uses multi-task learning to link drug targets and pathway activation, revealing interactions that guide tissue-specific combination therapies.

The paper identifies transcriptomic features of therapeutic targets that predict clinical success using Omic data and machine learning.

The paper predicts accurate binding modes of PFCAs in PPARγ using combined docking, MD simulations, and FEP+ to improve toxicity studies.

The paper introduces Decagon, a graph convolutional network that predicts polypharmacy side effects by modeling drug interactions and side effects as a multimodal graph, achieving up to 69% better performance than baselines.

The paper develops a multi-rank, multi-layered recommender system, ANTENNA, to predict drug-gene-disease associations and repurposes diazoxide as a potential anti-cancer therapy for triple-negative breast cancer.

This study compares three algorithms for identifying similar ligand-binding pockets in proteins and finds that G-LoSA performs best, while combining algorithms improves performance.

IMPPAT is a curated database of Indian medicinal plants, phytochemistry, and therapeutics, providing a resource for natural product-based drug discovery with 9596 phytochemicals and 1124 therapeutic uses.

The paper identifies a strong trend in protein-ligand interactions where proteins prefer to donate hydrogen bonds and ligands accept them, leading to the development of tools to predict and design ligand complexes.

The paper introduces a non-parametric method to measure discriminative power of drug-gene associations and identifies 128 new genomic markers across 61% of analyzed drugs.

The paper introduces Screenlamp, a toolkit for hypothesis-driven ligand screening, and demonstrates its use in discovering GPCR inhibitors for invasive species control.

The paper develops tools to compare and design optimized small molecule libraries using chemical and biological data, resulting in a more effective kinase inhibitor library.

The paper introduces Probe Miner, a resource that objectively assesses chemical probes using public data to evaluate over 1.8 million compounds against 2,220 human targets.

gene2drug prioritizes drugs based on their impact on transcriptional pathways of target genes, validated both computationally and experimentally for two drug repositioning cases.