I am a bioinformatics PhD candidate at the Institute of Bioinformatics, University of Georgia, studying gene duplication and lineage-specific gene evolution in Populus through comparative transcriptomics and phylogenetics. I expect to defend in Fall/Winter 2026.
Most recently I have been working on the robustness of genomic foundation models: parameter-efficient fine-tuning on DNA and RNA language models, and mechanistic interpretability to find where they fail and whether their confidence tracks real biology. It started as a side question in my dissertation, namely whether these models are useful on a non-model organism, and turned into its own line of work.
This fall I am also consulting part-time with the Computational Biology Consulting Group at the Institute of Bioinformatics, led by Dr. Casey Bergman.
Before UGA I did an MSc in Horticultural Crops Science at National Taiwan University, where I studied mycorrhiza-enhanced salt tolerance in rice and ran RNA-seq experiments on sex determination in bitter gourd.
Education
University of Georgia — PhD Candidate, Bioinformatics Expected December 2026
National Taiwan University — MSc, Horticultural Crops Science June 2019
National Taiwan University — BS, Agriculture June 2017
What I work with
Programming: Python · R · Bash/Shell · JavaScript · C/C++
Deep learning & foundation models: PyTorch, Hugging Face, genomic and RNA language models (Nucleotide Transformer, PlantCAD2, DNABERT2, AIDO.RNA, mRNA-FM, CodonFM), LoRA and parameter-efficient fine-tuning, embedding extraction, pooling, and frozen probing
Interpretability & robustness: per-layer linear probing, in-silico mutagenesis, per-position importance attribution, decision-depth analysis, black-box adversarial attacks and training, benchmark-validity auditing
Genomics & classical ML: RNA-seq pipelines (STAR, HISAT2, DESeq2, edgeR), 16S metagenomics (DADA2, QIIME2, SILVA, alpha/beta diversity, MaAsLin2), comparative transcriptomics, phylogenetics, genome-wide annotation, CRISPR gRNA design, XGBoost and SHAP
Engineering & deployment: Git/GitHub, Docker, CI/CD, Python packaging (PyPI), REST APIs, Snakemake, HPC/SLURM, agentic research workflows, Streamlit and R Shiny
Outside the dissertation
Much of what I build is meant to be handed to other people. The mentor–mentee matching platform I architected for the Taiwanese Young Researchers Association grew a program from 274 to 700+ annual participants and facilitated 1,876 researcher connections; the organization later partnered with Fulbright Taiwan. I organize the Institute of Bioinformatics seminar series at UGA, and I have led hackathon teams at HudsonAlpha HATCH (2024, 2025) and the UGA-Bayer Hackathon (2025).
Some of what I make is not serious. There is a DNA-to-audio sonification package, a sequence analysis library that speaks in internet slang, a research greenhouse rendered as a game of Snake, and a Reigns-style card game about surviving graduate school. It is all on the projects page.
Short bio
For conference programs and seminar introductions. Copy freely.
Chen Hsieh is a bioinformatics PhD candidate at the Institute of Bioinformatics, University of Georgia, where he studies gene duplication and lineage-specific gene evolution in Populus. His recent work examines the robustness of genomic foundation models, using constrained adversarial attacks and mechanistic interpretability to test whether these models have learned biology or their benchmarks. He holds an MSc in Horticultural Crops Science from National Taiwan University.
The compressed version is my resume, also available as a PDF. Longer versions: research, genomic AI, publications, teaching and community. A running log of what I’m working on week to week is my now page.