Computational Drug Discovery Methods

The paper introduces fingeRNAt, a tool for detecting non-covalent interactions in nucleic acid-ligand complexes and presents its applications for high-throughput analysis and similarity measurement.

This paper introduces retro drug design (RDD), an AI method to generate novel small molecule drugs with desired properties, and demonstrates its ability to produce compounds with biological activity and BBB permeability.

This paper introduces iGenSig, a genomic signature method for predicting cancer therapy responses using genome-wide sequencing data, showing strong cross-dataset performance and resilience to sequencing errors.

The paper uses machine learning on Tox21 data to identify protein features and pathways linked to compound toxicity, revealing biological mechanisms and cascading interactions.

This paper introduces a Deep Attention Neural Network framework for predicting drug-drug interactions, showing improved performance over existing methods.

This paper uses Deep Graph Networks to predict compound effects on phagocytosis and identifies 64 potential schizophrenia treatments from an existing drug library.

STRIFE improves fragment elaboration by incorporating pharmacophoric information from protein targets, outperforming existing methods in generating ligand-efficient structures.

This study uses multipartite network models to predict effective drug combinations and patient subclasses in acute myeloid leukemia using drug response data.

This paper introduces a graph neural network model for predicting molecule carcinogenicity, showing improved performance over existing methods.

The paper develops a graph convolutional network to screen compounds for SARS-CoV-2 entry inhibitors, identifying 16 active compounds from 170,000 virtual candidates.

The paper develops a knowledge network to prioritize drug mechanisms for repositioning, finding it can predict mechanisms with high accuracy.

The paper systematically explores human genome target spaces for protein and peptide drugs, identifying distinct characteristics and developing genome-wide target estimation models for each.

This paper identifies two natural compounds, Strophanthidin and Isopteropodin, as potential inhibitors of Brucella melitensis Methionyl-tRNA synthetase through in-silico screening and analysis.

The paper defines 481 conserved 3D ligand-binding motifs based on functional groups using an alignment-free method, applicable to drug-target interactions.

This study identifies three new potential inhibitors of hPGDS through computational modeling and molecular dynamics simulations, showing better interaction than existing compounds.

RPath prioritizes drugs for diseases by analyzing causal paths in knowledge graphs using drug-perturbed and disease-specific transcriptomic signatures, outperforming existing methods.

PepSeA provides tools for peptide sequence alignment and visualization to support lead optimization in drug discovery.

This study identifies an optimal strategy for drug repositioning using LINCS data and applies it to discover new therapeutics for liver cancer, showing homoharringtonine as a promising candidate.

This paper predicts effective drug combinations for treating early- and late-stage severe SARS-CoV-2 infection using an executable network model of host-virus interactions.

The paper introduces a deep learning framework, CIGER, for predicting gene expression rankings induced by chemicals and applies it to pancreatic cancer drug repurposing, showing improved performance over existing methods.

X-MOL uses large-scale pre-training on 1.1 billion molecules to improve molecular analysis tasks through fine-tuning, achieving state-of-the-art results in property prediction, reaction analysis, and molecule generation.

SynergyFinder Plus improves interpretation and annotation of drug combination screening data through updated methods and a web interface.

The paper presents a workflow for integrating diverse datasets to build a network pharmacology resource for COVID-19 research, resulting in a Neo4j graph database named Neo4COVID19.

This paper uses machine learning to predict disease-specific drug targets by integrating drug-induced expression profiles and gene perturbation data, finding significant enrichment of known targets.

The paper builds a knowledge graph analytics platform integrating LINCS and IDG to identify and prioritize novel drug targets for Parkinson's disease.

The paper introduces a non-parametric synergy model, Hand-GP, which outperforms existing parametric models in assessing drug and toxin combination effects.

The paper introduces a context-specific interaction framework to improve prediction of drug protein-protein interactions for unknown drug pathways, achieving higher accuracy than random methods.

The paper combines deep neural networks and Inductive Logic Programming to generate potential inhibitors for targets with limited data, using domain knowledge to improve molecule generation.

DenovoProfiling is a web server that provides visualization and functional profiling of de novo generated molecule libraries, offering tools for structural analysis, chemical space exploration, and target pathway identification.

FAME generates novel molecules with desired biological activity by learning from gene expression data using a conditional variational autoencoder and fragment-based generation.