Truong Son Hy

The paper introduces MutaStruct, a curated benchmark for evaluating mutation-induced structural changes, showing that overall structural fidelity does not guarantee accurate modeling of mutation effects.

PRIME-ISP predicts interaction sites across diverse biomolecular interactions using a hierarchical structural framework, outperforming sequence-based methods and matching specialized approaches without task-specific features.

This paper develops a machine learning approach to automate EPG signal segmentation and classification, achieving high accuracy in identifying insect feeding behaviors.

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

This paper introduces a Generative AI-assisted Virtual Screening Pipeline for drug repurposing that enables blind screening of protein targets without prior structural knowledge.

DiscoEPG is a Python package that automatically annotates insect EPG signals with improved accuracy using machine learning, offering tools for analysis and visualization.

LightMed improves medical image segmentation by using FFT-based processing to enhance robustness against noise and adversarial attacks.

This paper introduces a machine learning-guided directed evolution framework using large language models to improve protein design efficiency and effectiveness.

This paper constructs the first knowledge graph for Traditional Vietnamese Medicine (TVM) by adapting existing Traditional Chinese Medicine (TCM) resources and develops a benchmark for knowledge graph-based question-answering.

This study introduces ProteinReDiff, a diffusion-based model for redesigning ligand-binding proteins using sequence and ligand SMILES inputs, showing effectiveness in affinity and structural preservation.

This paper introduces LDE, a variational autoencoder-based method for protein sequence design that improves efficiency by prioritizing high-fitness mutants in latent space, outperforming baselines on eight tasks.

This paper introduces a symmetry-preserving multimodal pretraining framework for proteins, achieving competitive results in multiple tasks using a PyTorch implementation.