Research Interests

RNA Biology and Predictive Diagnostics

The driving force behind my research is to decode the regulatory complexity of RNA systems and translate this knowledge into smarter diagnostics and therapeutic strategies. As biomedical science shifts beyond protein-focused models, non-coding RNAs—particularly microRNAs (miRNAs)—have emerged as central players in gene regulation, cellular signaling, and disease mechanisms. My work is situated at this intersection of biology and computation, where molecules are transformed into predictive models.

Through academic and professional experiences, my focus has converged on interconnected areas: non-coding RNA biology, molecular diagnostics, cancer systems biology, and machine learning in bioinformatics. I am especially intrigued by miRNA–mRNA interactions as they can uncover regulatory networks that are both biologically significant and clinically relevant. While conventional approaches have identified miRNA targets through assays and sequence-based predictions, I see untapped potential in incorporating RNA structure, energy landscapes, and evolutionary conservation into predictive modeling.

Modeling miRNA–mRNA Interactions

A major strand of my research is the development of computational pipelines that can reliably predict and prioritize miRNA binding across coding and non-coding RNAs. Current tools often emphasize seed-region complementarity, but growing evidence highlights the importance of RNA folding, accessibility, GC content, and local energy stability (∆G) in shaping binding outcomes.

My aim is to embed these features into deep learning frameworks—such as convolutional neural networks (CNNs) and transformer models—to capture the layered relationship between RNA structure and function. I am particularly drawn to conserved RNA motifs at regulatory hotspots, including UTRs, pseudoknots, and stem-loop structures. By applying attention mechanisms, I hope to interpret which structural or thermodynamic cues most strongly influence binding predictions.

One application of this approach is in structured viral RNAs, such as the HIV Rev Response Element (RRE), where host miRNAs may influence viral replication. More broadly, I seek to extend these methods to cancer-associated transcripts, especially in oncogenic and tumor suppressor pathways, where miRNA dysregulation is well-documented but not fully mapped at the systems level.

Multi-Omics Biomarker Discovery in Cancer

Another key interest of mine is multi-omics integration, where transcriptomic, genomic, and proteomic data are combined to construct richer models of disease. Although omics datasets are increasingly abundant, extracting meaningful biological insights remains a challenge. I am particularly interested in dimensionality reduction, graph-based neural networks, and Bayesian inference methods that can uncover functional modules and disease subtypes from complex data.

In cancer diagnostics, I envision pipelines that jointly analyze miRNA expression, gene mutations, splicing events, and proteomic profiles to: classify cancer subtypes, predict progression, anticipate drug response, and identify non-invasive biomarkers for early detection. To achieve this, I plan to use approaches such as Weighted Gene Co-expression Network Analysis (WGCNA), autoencoders, and integrated feature selection methods that prioritize features by biological relevance rather than frequency alone.

My long-term goal is to move from static biomarker lists toward adaptive diagnostic systems that continuously learn from new data.

RNA Structure and Molecular Dynamics

RNA function is shaped not only by sequence and interaction partners but also by its structural dynamics in the cell. My third area of focus is structure-based modeling of RNA molecules and RNA–RNA / RNA–protein complexes, with particular emphasis on miRNA–target duplexes.

Through molecular dynamics (MD) simulations and docking tools such as GROMACS, YASARA, and AutoDock, I aim to study how miRNA binding reshapes RNA conformation, accessibility, and downstream signaling. Coupled with free energy landscape analysis, these simulations can reveal the thermodynamic and kinetic feasibility of predicted interactions. This strategy also supports the rational design of therapeutic RNAs, including synthetic miRNAs, antagomiRs, aptamer-based biosensors, and RNA decoys. Integrating MD outputs into deep learning models could further improve prediction accuracy, enabling structure-aware AI systems to rank interaction strength across biological contexts.

Precision Public Health and AI-Driven Diagnostics

Beyond molecular modeling, I am equally invested in translating computational biology into practical diagnostic tools. I believe that AI-powered diagnostic platforms, trained on omics datasets, can enable scalable and decentralized healthcare solutions—particularly in resource-limited environments.

My broader vision is to design diagnostic workflows that can run on portable devices (e.g., smartphone-based qPCR analysis), leverage cloud-based prediction engines for RNA biomarker panels, and adapt dynamically to new health threats through continuous learning. Potential applications include RNA biomarker panels for early cancer detection, miRNA signatures for drug resistance profiling, and field-ready kits that combine RNA biosensors with AI-driven interpretation.