Wellcome Sanger Institute, Cambridge, UK
dj16@sanger.ac.uk
Research Interest
•
Multi-modal learning of biomedical images and molecular profile
•
Regulatory sequence generation
Research Vision
“What I cannot create, I do not understand”
- Richard Feynman
I was drawn to computational biology by the idea that DNA is the ultimate code: a sequence in which both code and the codebook are deeply entangled. My research ambition is to fully decode this language and build models that can simulate biological systems from sequence.
About me
I am an AI for Science researcher at the Lotfollahi Lab, Wellcome Sanger Institute, working at the intersection of machine learning and biomedical science.
My background in bioinformatics and biochemistry shapes how I approach this space — not by fitting biological data into existing AI frameworks, but by asking what a problem actually requires before reaching for a method. I focus on translating the complexity of biological and medical questions into well-defined computational problems, and building the models to solve them.
News
•
2026.06: SIGMMA accepted to ICML FM4LS/SD4H workshop. A journal version coming soon!
SIGMMA accepted to ICML FM4LS/SD4H workshop. A journal version coming soon!•
2024.10: Joined the Wellcome Sanger Institute, one of the world’s leading genomics research institutes
Joined the Wellcome Sanger Institute, one of the world’s leading genomics research institutes Professional Experience
Senior Data Scientist - Wellcome Sanger Institute, UK
2024.10 - Present
- Led the development of SIGMMA, a hierarchical contrastive learning framework that aligns H&E histopathology with spatial transcriptomics (ICML 2026 Workshop; journal version in preparation)
- Designed a generative framework for de novo design of regulatory DNA sequences, using property-guided generation to reach out-of-distribution objectives within a lab-in-the-loop design cycle
Research Intern - LG AI Research, Korea
2024.1 - 2024.4
- Contributed to development of a self-supervised foundation model for H&E histopathology images, ExaonePath
- Built a scalable and reproducible pipeline for multi-omics data preprocessing
Education
Ph.D in Interdisciplinary Program of Bioinformatics
Seoul National University, Korea | Graduated in 2024.08
•
Thesis: Computational modeling of molecular interactions for biomarker discovery using single/multi-omics data
•
Published 3 first-authored and 10 co-authored papers.
•
Developed GOAT, a graph neural network for multi-omics biomarker discovery in eosinophilic asthma (Bioinformatics, 2023)
•
Developed RNA-seq quantification and normalization pipelines as part of the International Cancer Genome Consortium – Accelerating Research in Genomic Oncology (ICGC-ARGO), making standardized preprocessing accessible to the broader research community.
B.S. in Biochemistry
Yonsei University, Korea | Graduated in 2018.08
Technical Skills
•
Programming: Python, R, Shell, LaTeX
•
ML / Deep Learning: PyTorch, Torch geometric, DGL, scikit-learn
•
Workflow & Reproducibility: Nextflow, Snakemake, Docker, Git
Awards & Achievements
Invited talk, National Heart & Lung Institute, Imperial College London – 2023•
"Deep graph attention model for multi-omics integration discovers network biomarkers for eosinophilic asthma subtype”, hosted by Prof. Kian Fan Chung
National Excellence Scholarship (Natural Sciences and Engineering) – 2016•
Selected as the sole recipient in my department for a merit-based government scholarship
Languages
•
English, Full professional proficiency
•
Korean, Native
•
Spanish, Conversational