Biology subjects

Nguyen, D. D.

Publications and source records attributed to Nguyen, D. D..

4 recordsLinked to original sources

bioRxiv2026

Multi-level, multi-body atomic interaction graphs for machine learning-based prediction of protein-ligand binding energies

Accurate prediction of binding affinity is crucial for rational drug design and discovery. Traditional computational methods often rely on complex scoring functions that incorporate a multitude of physical and chemical descriptors, leading to high computational demands and sometimes limited generalizability. In this work, we propose a novel scoring function that models multi-level, multi-body atomic interactions using graph-based representations. Our method constructs comprehensive interaction graphs that incorporate both pairwise and triplet-wise atomic features that help capture cooperative spatial patterns essential for binding affinity prediction. By employing a feature fusion strategy, GMI-Score maintains model simplicity while enhancing accuracy. Extensive evaluation across multiple datasets, such as PDBbind v2013, PDBbind v2016, PDBbind v2020, CSAR-NRC-HiQ, and PDBbind-Redocked, demonstrates that our model consistently outperforms state-of-the-art scoring functions, achieving Pearson correlation coefficients up to 0.877. Furthermore, it retains strong predictive power under strict data leakage controls and realistic docking conditions to high-light its robustness and generalizability. Scientific ContributionIn this study, we present a scoring methodology that systematically captures higher-order atomic interactions within a unified graph framework, making a conceptual shift in cheminformatics scoring functions. Its consistent outperformances of existing methods and strong validity under redocked and withheld atascenarios demonstrate its utility for broad-scale molecular modeling applications and open heminformaticsworkflows.

bioinformatics↗

bioRxiv2025

When does additional information improve accuracy of RNA secondary structure prediction?

The secondary structure of an RNA sequence plays an important role in determining its function, and accurate prediction of the structure is still a major goal in computational biology. Improvements in the prediction accuracy of the secondary structure can be achieved via auxiliary information. In this paper, we study features based on suboptimal formations competing with the minimum-free energy formation and investigate their role in determining the improvement of accuracy via auxiliary information, which we call directability. Here, we introduce a similarity measure among competing substructures called profiles. Then, we present an n-dimensional representation of the profiles which allows the use of topological data analysis (i.e., persistence landscapes) to obtain different metrics that represent topological features. Then, we built random forest classifiers using these novel features. We show how the similarity feature is more important for classifiers trained on sequences with similar structures while the topological features are more important for classifiers trained on sequences with dissimilar structures. We perform extensive testing on two sets of RNA sequences where we studied the sensitivity of the classification accuracy and their feature importance.

bioinformatics↗

bioRxiv2023

The Interplay between Mutagenesis and Extrachromosomal DNA Shapes Urothelial Cancer Evolution

Advanced urothelial cancer is a frequently lethal disease characterized by marked genetic heterogeneity. In this study, we investigate the evolution of the genomic signatures caused by endogenous and external mutagenic stimuli and their interplay with complex structural variants. We superimposed mutational signatures and phylogenetic analyses of matched serial tumors from patients with urothelial cancer to define the evolutionary patterns of these processes. We show that APOBEC3-induced mutations are clonal and early, whereas mutational bursts comprising hundreds of late subclonal mutations are induced by chemotherapy. Using a novel genome graph computational paradigm, we observed frequent circular high copy-number amplicons characteristic of extrachromosomal DNA (ecDNA) involving double-minutes, breakage-fusion-bridge, and tyfonas events. We characterized the distinct temporal patterns of APOBEC3 mutations and chemotherapy-induced mutations within ecDNA, gaining new insights into the timing of these events relative to ecDNA biogenesis. Finally, we discovered that most CCND1 amplifications in urothelial cancer arise within circular ecDNA amplicons. These CCND1 ecDNA amplification events persisted and increased in complexity incorporating additional DNA segments potentially contributing selective fitness advantage to the evolution of treatment resistance. Our findings define fundamental mechanisms driving urothelial cancer evolution and have therapeutic implications for treating this disease.

genomics↗

bioRxiv2020

Machine intelligence design of 2019-nCoV drugs

Wuhan coronavirus, called 2019-nCoV, is a newly emerged virus that infected more than 9692 people and leads to more than 213 fatalities by January 30, 2020. Currently, there is no effective treatment for this epidemic. However, the viral protease of a coronavirus is well-known to be essential for its replication and thus is an effective drug target. Fortunately, the sequence identity of the 2019-nCoV protease and that of severe-acute respiratory syndrome virus (SARS-CoV) is as high as 96.1%. We show that the protease inhibitor binding sites of 2019-nCoV and SARS-CoV are almost identical, which means all potential anti-SARS-CoV chemotherapies are also potential 2019-nCoV drugs. Here, we report a family of potential 2019-nCoV drugs generated by a machine intelligence-based generative network complex (GNC). The potential effectiveness of treating 2019-nCoV by using some existing HIV drugs is also analyzed.

bioinformatics↗