The central motivation behind my research interests lies in understanding the regulatory complexity of RNA-based systems and their applications in developing intelligent diagnostics and therapeutic solutions. As biomedical science moves beyond protein-centric frameworks, the roles of non-coding RNAs, especially microRNAs (miRNAs), have become pivotal in shaping gene expression, cellular response, and disease progression. My research is positioned at this evolving frontier—where biology meets computation, and where molecules become models.
My academic and professional exposure has crystallized my focus into several interlinked domains: non-coding RNA biology, molecular diagnostics, cancer systems biology, and machine learning applications in bioinformatics. I am particularly fascinated by the potential of miRNA–mRNA interactions to reveal regulatory circuits that are not only biologically meaningful but also clinically actionable. While traditional methods have illuminated miRNA targets through experimental assays and sequence-based tools, I believe there is significant untapped potential in harnessing RNA structure, energy profiles, and evolutionary conservation as predictive features in modeling these interactions.
A substantial portion of my research interest lies in developing computational frameworks that accurately predict and rank miRNA binding sites across coding and non-coding transcripts. While most current tools emphasize base-pairing complementarity at the seed region, emerging evidence shows that RNA secondary structure, accessibility, GC content, and local folding energy (∆G) significantly modulate binding efficiency and biological outcomes.
I aim to integrate these features into deep learning-based architectures, such as convolutional neural networks (CNNs) and transformer models, to capture hierarchical relationships between structure and function. In this context, I am particularly interested in RNA structural elements that are conserved across species and enriched at key regulatory loci, such as UTRs, pseudoknots, and stem-loop motifs. Additionally, I hope to apply attention mechanisms within these networks to visualize and interpret what structural or thermodynamic signals drive binding prediction.
One specific area of application is in studying structured viral RNAs, such as the Rev Response Element (RRE) in HIV, where host miRNAs may regulate viral replication. However, my broader goal is to apply these methods to human cancer-related transcripts, particularly in oncogenic and tumor suppressor pathways, where miRNA deregulation has been extensively implicated but not fully mapped at the systems level.
My second major interest lies in multi-omics data integration, where transcriptomic, genomic, and proteomic layers are fused to build context-rich models of disease. While omics data is increasingly available, its biological interpretation remains a challenge. I am particularly drawn to dimensionality reduction techniques, graph-based neural networks, and Bayesian inference frameworks that allow researchers to infer functional modules and disease subtypes from complex datasets.
For example, in cancer diagnostics, I envision building multi-modal pipelines where miRNA expression, gene mutations, alternative splicing events, and proteomic signatures can be jointly analyzed to:
-
Predict cancer subtypes and progression stages,
-
Forecast patient-specific drug response, and
-
Identify non-invasive biomarkers for early detection.
To achieve this, I aim to employ tools such as Weighted Gene Co-expression Network Analysis (WGCNA), autoencoders, and integrated feature selection algorithms that prioritize molecular features not just by frequency but by biological impact and redundancy minimization.
Ultimately, I want to transition from static biomarker lists to adaptive diagnostic frameworks that learn from new data, adapt to patient heterogeneity, and provide continuous refinement of predictive outputs.
The functional role of RNA is not only defined by its sequence or interaction network but also by its structural dynamics in the cellular context. Therefore, my third area of research interest is in the structure-based modeling of RNA molecules and RNA–RNA / RNA–protein complexes, particularly in miRNA-target duplexes.
Using molecular dynamics (MD) simulations and molecular docking tools like GROMACS, YASARA, and AutoDock, I hope to investigate how miRNA binding alters target RNA conformation, accessibility, and downstream signaling. These simulations, paired with free energy landscape analyses, can provide insight into the thermodynamic feasibility and kinetic stability of candidate interactions predicted in silico.
Additionally, this approach can support rational design of therapeutic RNAs, such as:
-
Synthetic miRNAs or antagomiRs,
-
Aptamer-based detection systems,
-
RNA decoys or sponges for competitive inhibition.
The integration of MD simulation results into deep learning models may further enhance prediction accuracy, enabling structure-aware ML models capable of ranking interaction strength across diverse biological contexts.
Beyond molecular modeling, I am equally passionate about the translational application of computational biology to real-world diagnostics. In particular, I believe that AI-assisted diagnostic platforms, powered by machine learning models trained on omics data, hold the key to scalable, decentralized health technologies—especially in low-resource settings.
My broader interest lies in designing diagnostic workflows that can:
-
Be run on portable devices (e.g., smartphone-based analysis of qPCR results),
-
Leverage cloud-based prediction tools for RNA biomarker panels, and
-
Adapt to emerging health threats through continuous model learning.
Examples of such platforms may include:
-
RNA biomarker panels for early cancer detection,
-
miRNA panels for drug resistance profiling, or
-
Field-adaptable kits integrating RNA-based biosensors with AI-powered interpretation.