Current focus
Computational genomics and TE data
- Transposable element annotation
- TE taxonomy construction
- ML-ready biological dataset curation
- Computational genomics
- Biological sequence modeling
- Multimodal TE annotation model concept
Computational biology · Computational genomics
Nina Deng
Computational Genomics & Biological AI
I work on transposable element annotation and ML-ready genomics datasets, and I am developing a multimodal TE annotation model concept. I am expanding toward physics-informed biological foundation models and virtual cell systems.
Early-career researcher with a background in math-biology, computational genomics, data modeling, and visual communication.
01 · Research focus
A concrete foundation,
with a clear direction.
My current work is grounded in computational genomics and TE annotation. From this base, I am exploring how biological foundation models can become more mechanistically grounded, physically constrained, and useful for reasoning across biological scales.
Expanding toward
Mechanistically grounded biological AI
- Genomics foundation models
- Biological foundation models
- Physics-informed biological AI
- Virtual cell systems
- Dynamical-systems thinking
- Mechanistically grounded representation learning
02 · Questions I’m exploring
Questions before claims.
A few questions currently shaping my research direction:
- 01
How can genomic sequence modeling connect to cell-level biological representation learning?
- 02
How can biological foundation models become more mechanistically grounded rather than purely predictive?
- 03
What dynamical, physical, or energy-based constraints are meaningful for biological AI?
- 04
How can messy biological annotation systems become model-ready, interpretable structures?
- 05
How can sequence, domain, structure, and taxonomy evidence be integrated into annotation models?
03 · Current research projects
Research assets
and active directions.
Three connected areas: building sound data foundations, developing an interpretable annotation framework, and defining the next research bridge.
Harmonizing TE labels and metadata into a structured dataset for downstream modeling.
TE Dataset Curation & Taxonomy Harmonization
Curating and organizing transposable element consensus-sequence data from RepBase and Dfam-derived sources into a cleaner, ML-ready format. The work includes sequence deduplication, label-conflict review, ontology-aware taxonomy mapping, source metadata organization, and hierarchical label harmonization.
Why it matters: This work turns heterogeneous biological database records into a more structured, traceable resource for downstream modeling.
Conceptual architecture for multimodal TE annotation combining sequence, domain, structural, and taxonomy evidence.
Multimodal TE Annotation Model Concept
Developing a research framework for library-independent transposable element annotation that would integrate genomic sequence representations, protein-domain evidence, structural signals, and hierarchical TE taxonomy. The aim is to explore a path beyond purely library-based annotation toward interpretable, multimodal biological sequence modeling.
Current scope: This is a conceptual research framework and developing set of modeling ideas—not a completed production system.
A research direction connecting computational genomics to physics-informed biological foundation models.
Toward Mechanistically Grounded Biological AI
Exploring how biological foundation models might become more mechanistically grounded through dynamical-systems thinking, energy-dissipation principles, and cross-scale biological representation learning. This direction connects computational genomics with future-facing questions in physics-informed biological AI and virtual cell systems.
Current scope: A future-facing research direction being developed from the concrete base of TE data and genomic sequence modeling.
04 · Cross-domain evidence
One underlying skill:
structuring complexity.
Earlier work that shaped how I model data, communicate complex ideas, and organize projects across research, design, and public-facing contexts.
Data Modeling &
Analytical Thinking
Turning messy data and abstract problems into structured analytical workflows.
- Analyzed rare-earth-element data from Appalachian Mountain samples using Python during undergraduate research at Miami University.
- Presented geochemical and mineralogical analysis at a student research symposium.
- Led algorithm development for shredded-paper reconstruction using textual features, hierarchical clustering, and a Traveling Salesman Problem formulation.
- Worked with BI and reporting workflows using Python, Excel, SQL, and PowerPoint.
Research relevance Biological datasets are similarly heterogeneous and sensitive to modeling assumptions.
Visual Communication &
Data Storytelling
Making complex and public-facing information visible, readable, and persuasive.
- Created data visualizations and presentation materials using MATLAB, Python, and PowerPoint.
- Designed public-facing articles and posters for large-scale cycling events.
- Used Photoshop, Procreate, and visual design workflows to communicate beyond plain text.
- Translated data and concepts into presentation-ready visual formats.
Research relevance Supports research presentations, proposal writing, visual explanation, and outreach.
Leadership &
Outreach
Coordinating small analytical projects and contributing to public-interest communication.
- Served as team leader for a business-intelligence and win-rate analysis project using Python and PowerPoint.
- Led algorithm development within a mathematical modeling project.
- Contributed visual design and volunteer support for animal-protection and adoption events.
- Built experience coordinating tasks, presenting results, and supporting public-facing work.
Research relevance Supports collaborative research, lab communication, outreach, and project execution.
05 · Selected visual work
Visual & communication samples.
A small selection of placeholders for visual communication, data storytelling, and public-facing design work. Final images can replace these without changing the layout.
06 · Education
The Ohio State University
B.S. Mathematics, Biology Track
Research and coursework in computational biology, molecular genetics, statistics, mathematics, and genomics.
Earlier training included mathematics, modeling, and data analysis.
07 · Contact
Let’s talk about biological data, models, and the questions between them.
Open to research conversations, research-assistant opportunities, and collaborations in computational genomics, AI for biology, biological foundation models, and physics-informed biological modeling.
A PDF version of the research memo will be added before public launch.