Pillar I · Curriculum
Foundations
An evergreen, first-principles path from protein structure through machine learning and protein language models to generative drug discovery.
8 of 40 articles published20%
Research Officer · National Research Council Canada
Junan Lin
Physicist turned computational biologist. I build deep-learning models that characterize proteins and design new drug molecules, building on a Ph.D. and postdoc in quantum information and quantum algorithms.
- Now
- Deep learning for protein characterization and drug design at NRC Canada
- Training
- Ph.D. Physics (Quantum Information), University of Waterloo
- First-author papers
- Scientific Reports · Phys. Rev. A · Phys. Rev. Research · New J. Phys.
- Honours
- NSERC PGS-D and CGS-M scholarships
The site in two pillars
Pillar II · Research stream
Latest Developments
What is new in computational biology and drug discovery: paper breakdowns, intermediate research results, and tools or visualizations worth sharing.
- Paper breakdownsImportant new papers, read closely, with the math and the caveats.
- Research notesIntermediate results and lessons from my own projects.
- Tools & visualizationsInteractive figures and code worth sharing.
First entries are in preparation.
Research
Selected work
NRC Canada · 2024–present
AI for drug design
BiRLNN, a bidirectional RNN trained with reinforcement learning for constrained molecular generation (Scientific Reports, 2026). Current work: antibody affinity maturation with cross-attention over protein-language-model embeddings, and conditional 3D ligand design with flow matching.
University of Toronto · 2023–2024
Quantum algorithms for chemistry
Rigorous, computable bounds on the overlap between a trial wavefunction and Hamiltonian eigenstates, which turn into performance guarantees for quantum phase estimation.
University of Waterloo · 2016–2023
Quantum error characterization
Characterizing state-preparation and measurement errors independently of gate errors, and suppressing them with algorithmic cooling (New J. Phys. 2019; Phys. Rev. Research 2021; Phys. Rev. A 2022, 2024).
About
About me
I'm a Research Officer at the National Research Council Canada, where I work on applying deep learning to protein characterization and de novo small molecule drug design. My current projects include generative AI frameworks for ligand design, protein language model embeddings for antibody affinity maturation, and reinforcement learning for constrained molecular generation.
Before this, I was a postdoctoral fellow at the University of Toronto working on quantum algorithms for quantum chemistry, and before that I completed my Ph.D. in Physics (Quantum Information) at the University of Waterloo under Prof. Raymond Laflamme. My doctoral work focused on characterizing and mitigating errors in quantum processors — a problem that blends quantum information theory, physics, and statistics.
This blog is an ongoing effort to do two things at once: educate myself on the essential foundations of structural and computational biology, and then pass that knowledge on to younger researchers who want to enter the field. I have always enjoyed teaching, and I have found that I am at my best when I break complex ideas into understandable pieces. In practice, this blog is also a test of something I deeply believe: teaching is one of the best ways to learn.
Throughout my academic and professional career, I have tried to follow my curiosity toward problems that can create tangible social impact. When I started my current role and began my first project, I was genuinely amazed by the pace of progress in this field, and I believe much more is still to come. I also think the next phase will require more passionate researchers who can tackle harder problems with the help of AI. I hope this blog can be one small contribution to that effort, and a source of motivation for people considering this path.
Selected publications (* = first or co-first author)
Full list on Google Scholar →- 1.Bidirectional reinforcement learning neural network for constrained molecular design
*Junan Lin, Mohammad Sajjad Ghaemi, Jiří Hostaš, Anguang Hu, Hang Hu, Hsu Kiang Ooi
Scientific Reports 16, 3393 (2026)
- 2.VNFlow: Integration of Variational Autoencoders and Normalizing Flows for Novel Molecular Design
*Jiří Hostaš, Mohammad Sajjad Ghaemi, Anguang Hu, Hang Hu, Junan Lin, Hsu Kiang Ooi
J Cheminform 17, 161 (2025)
- 3.Bounds on a Wavefunction Overlap with Hamiltonian Eigen-states: Performance Guarantees for the Quantum Phase Estimation Algorithm
*Junan Lin, Artur F. Izmaylov
arXiv:2503.12224 (2025)
- 4.Thermodynamic Analysis of Algorithmic Cooling Protocols: Efficiency Metrics and Improved Designs
*Junan Lin, *Nayeli Azucena Rodríguez-Briones, Eduardo Martín-Martínez, Raymond Laflamme
Physical Review A 110, 022215 (2024)
- 5.Algorithmic cooling for resolving state preparation and measurement errors in quantum computing
Raymond Laflamme, *Junan Lin, Tal Mor
Physical Review A 106 (1), 012439 (2022)
- 6.Independent state and measurement characterization for quantum computers
*Junan Lin, Joel J Wallman, Ian Hincks, Raymond Laflamme
Physical Review Research 3 (3), 033285 (2021)
- 7.On the freedom in representing quantum operations
*Junan Lin, *Brandon Buonacorsi, Raymond Laflamme, Joel J Wallman
New Journal of Physics 21 (2), 023006 (2019)