Computational Drug Discovery Methods

Overcoming Topology Bias and Cold-Start Limita-tions in Drug Repurposing: A Clinical-Outcome-Aligned LLM Framework

This paper introduces a framework that improves drug repurposing by aligning LLM reasoning with clinical outcomes, overcoming topology bias and cold-start limitations.

Large Scale Cell Painting Guided Compound Selection Reveals Activity Cliffs and Functional Relationships

This paper introduces a transfer learning method for compound preselection using phenotypic profiling, revealing activity cliffs and functional relationships across 112,480 compounds.

Enhancing CYP450-Ligand Binding Predictions: A Comparative Analysis of Ligand-Based and Hybrid Machine Learning Models

This paper compares ligand-based and hybrid machine learning models for predicting CYP450 ligand binding, finding the hybrid model outperforms existing methods in accuracy.

COMIC: Explainable Drug Repurposing via Contrastive Masking for Interpretable Connections

COMIC improves drug repurposing prediction with explainable pathways, achieving 9.55% better performance than existing methods on a large knowledge graph.

Multimodal Bonds Reconstruction Towards Generative Molecular Design

YuelBond improves bond reconstruction in molecular design by handling 3D structures, distorted compounds, and 2D graphs with high accuracy.

Deep contrastive learning enables genome-wide virtual screening

DrugCLIP, a contrastive dense-retrieval model for pockets and molecules, beats docking on DUD-E and LIT-PCBA and screened about 10,000 human AlphaFold2 proteins in 24 hours.

Graph-based Drug Decomposition for Anticancer Response Modeling

This paper models ex vivo drug response in pediatric AML by decomposing drug molecules into structured features and using graph-based methods to identify key molecular predictors of cellular viability.

BindFlow: a free, user-friendly pipeline for absolute binding free energy calculations using free energy perturbation or MM(PB/GB)SA

BindFlow is a Python-based tool for absolute binding free energy calculations using FEP or MM(PB/GB)SA, showing performance comparable to gold standards with lower computational cost for some systems.

AI-Assisted Discovery and Optimization of Small Molecule TREM2 Agonists with Functional Microglial Activity

Deep Docking virtual screening of over five million compounds and structure-activity optimization yielded T2M-010, a submicromolar small-molecule TREM2 agonist that induces SYK phosphorylation and microglial phagocytosis.

From Latent Manifolds to Functional Probes: An Interpretable, Kinome-Scale Generative Machine Learning Framework for Family-Targeted Kinase Inhibitor Design

This paper introduces a generative framework for designing SRC kinase inhibitors, showing that inhibitors organize into a low-dimensional manifold and highlighting representation gaps in SMILES-based generation.

Discovery of potent oligopeptides for various metabolic diseases using deep learning

This paper introduces Deepeptide, a deep learning pipeline that identifies therapeutic oligopeptides for metabolic diseases, finding 62% of candidates show significant in vitro bioactivity comparable to first-line drugs.

Protein Language Models and Structure-Based Machine Learning for Prediction of Allosteric Binding Sites in Protein Kinases: An Explainable AI Framework Grounded in Energy Landscape-Encoded Frustration

This paper compares sequence-based and structure-based methods for predicting allosteric binding sites in protein kinases, finding that sequence-based models perform poorly on allosteric sites despite good performance on orthosteric ones.

Delta PSA: A New Metric for Conformational Dynamics Underlying Macrocyclic Peptide Permeability

The paper introduces Delta PSA, a metric for predicting macrocyclic peptide permeability using MD simulations, achieving lower MSE than existing methods.

Learning a PRECISE language for small-molecule binding

PRECISE improves virtual screening by combining quantized drug representations with 3D target structures to predict binding sites and enable billion-scale screening.

PRADA-DTI: A Prototype-Retrieval Augmented Domain-Adaptation Framework for Drug-Target Interaction Prediction

PRADA-DTI improves drug-target interaction prediction by enabling continual learning from new protein domains with minimal parameter updates and reduced forgetting.

TubercuProbe: A Cross-Attention Graph-Sequence Model for Cross-Species Chemoproteomic Discovery in Mycobacterium tuberculosis

TubercuProbe prioritizes compound-protein interactions in Mycobacterium tuberculosis using a cross-attention graph-sequence model trained on human chemoproteomic data.

Phenotype-Guided In Silico Molecular Generation Using Large Language Models

GEMGen generates small molecules from transcriptomic data to phenocopy gene knockdown effects and identify novel inhibitors.

NetMedGPT - A network medicine foundation model for extensive disease mechanism mining and drug repurposing

NetMedGPT is a transformer-based foundation model trained on biomedical knowledge graphs to predict drug-disease associations and support drug repurposing, outperforming specialized models in multiple tasks.

ASSESSMENT OF ALPHAFOLD PROTEIN MODELS FOR SMALL-MOLECULE LIGAND DOCKING

This study evaluates AlphaFold protein models for ligand docking, showing they outperform unbound structures and match ligand-bound structures in fragment docking.

Characterizing Clinical Toxicity in Cancer Combination Therapies

This study examines if synergy scores and toxicity metrics correlate with known adverse drug interactions, finding significant limitations in using them as penalties for toxicity in synergy prediction.

Structure-Based Virtual Screening Identifies TREM2-Targeted Small Molecules that Enhance Microglial Phagocytosis

This paper identifies a small molecule TREM2 modulator, EN020, that enhances microglial phagocytosis better than a known agonist.

Accelerating ligand discovery by combining Bayesian optimization with MMGBSA-based binding affinity calculations

The paper combines Bayesian optimization with MMGBSA-based binding affinity calculations to improve ligand discovery by enhancing compound ranking and recovery rates.

PocketBagger: Generalizable pocket druggability prediction via positive-unlabeled learning

PocketBagger predicts druggable protein pockets using positive-unlabeled learning, achieving high recall and generalizability across protein families.

Daclatasvir, a symmetric drug for an anti-symmetric target

This study identifies daclatasvir, an FDA-approved antiviral, as a potential PD-L1 inhibitor through computational methods, showing it binds stably to the PD-L1 homodimer.

Identification of drug candidates against glioblastoma with machine learning and high-throughput screening of heterogeneous cellular models

This study identifies three compounds targeting glioblastoma stem cells using machine learning and high-throughput screening of patient-derived models.

Enumerating the chemical exposome using in-silico transformation analysis : an example using insecticides

The paper demonstrates an in-silico workflow to enumerate chemical transformation products of insecticides using machine learning and chemical data resources, identifying 19,392 unique products.

Candidate NAMPT modulators from traditional African, Chinese, and Russian medicinal plants

The paper identifies 17 potential NAMPT modulators from traditional medicinal plants and FDA libraries through in-silico screening and molecular docking validation.

Teaching Diffusion Models Physics: Reinforcement Learning for Physically Valid Diffusion-Based Docking

This paper improves diffusion-based molecular docking by using reinforcement learning to enforce physical validity, enhancing pose accuracy without increasing computational cost.

Sharing Patterns Between Proteins and Exploring Possible Drug Interactions

The paper identifies shared amino acid patterns in cancer-related proteins and explores their potential for drug interaction analysis, focusing on ALDH2 and disulfiram.

Leveraging Uncertainty Estimates for Drug Response Prediction in Cancer Cell Lines

This paper benchmarks uncertainty-aware models for drug response prediction in cancer cell lines, finding ensemble-based methods better detect distribution shifts while distributional models reduce prediction errors in high-confidence subsets.