about
Hi, I’m Young Su, a Molecular Biophysics PhD student at UC San Diego.
Welcome to my notebook. Let me explain what this is and why I made it.
my backstory (college - 2022)
My PhD research is focused on using deep learning to better understand the behavior of proteins, like predicting whether two proteins will interact. But when I started my PhD in 2022, I didn’t think I’d be working in this area. I knew I wanted do computational research1, but I never thought I’d be working on deep learning.
For my Bachelor’s, I majored in Chemistry at Pomona College. Back then, I was running molecular docking and molecular dynamics (MD) simulations, trying to see whether small molecules would bind to the SARS-CoV2 main protease. I applied to UCSD because I wanted to join the Amaro Lab, where they do cutting-edge MD with lots of biomedical applications.
So, how did I end up at the Wang Lab? The Chemistry and Biochemistry program at UCSD requires first-year PhD students do 3 rotations. There was another lab doing MD, so I had my winter and spring rotations set up but I couldn’t figure out what to do for my first fall rotation. My first-year advisor at the time recommended the Wang Lab. I didn’t know anything about the lab, but it was the only other lab doing computational research, so I signed up.
the rotation
I began my rotation with the Wang Lab in the Fall of 2022. I got connected with a research scientist who needed help with a project they were finishing up. In this project2, they trained a model to predict whether an antibody would bind to PD-L1, and used an autoencoder to represent the antibodies. My job was to see how antibody language models (AntiBERTy, AbLang) representations would perform when used as input to the model.
It was my first time learning about language models, first time hearing about PyTorch, heck, I didn’t even know how to use a GPU. I struggled with the most basic things (using SLURM, tensors, conda environments). It took me the entire month, to do the simple task of generating embeddings. But in the end, I was surprised by how much I enjoyed this rotation.
why i joined
I always assumed AI and machine learning was out of reach for me given my chemistry background. I didn’t take any CS classes in college and while I was familar with Python, this rotation taught me I knew way less about Python than I thought I did. I knew I’d be completely out of my depth if I joined this lab. Yet, the research really fascinated me. I got a glimpse at how AI was being used to develop anti-cancer antibodies.
Having lost my mom to cancer, this project was especially meaningful to me. Although I’ve always been interested in research with biomedical applications, it was the first time I got to see how AI could be transformative in this field. I began to believe, as I still do, that AI will have a meaningful impact in medicine. I knew I lacked experience. I knew I lacked the knowledge. Yet, I felt a calling and wanted nothing more than to contribute to this field.
insecurities & imposter syndrome
After joining the lab, I began my journey into the world of deep learning for proteins. I knew I needed to catch up. I tried to learn about deep learning from anywhere I could–classes at UCSD, watching lectures on YouTube, talking to ChatGPT, reading Twitter threads, etc.
Still, I felt like an imposter. I didn’t want to tell people I was working on machine learning for proteins, because I was afraid they would ask the obvious, “what do you know about machine learning?” So fear was my motivation–I thought if I tried really hard, I could catch up before anyone noticed I didn’t belong. But the more I learned, the more I realized how little I know.
taking ownership
Looking back, it was extremely naive, and maybe even arrogant, of me to think this way, to think that I could quietly catch up on everything. It’s been three years since I started this journey and I have about two years left before I graduate. I realized something. I could spend the remaining two years, hiding–hiding what I don’t know behind surface-level intuition and flying under the radar.
But I realized I’d rather spend my last years being honest with myself. It’s a little scary, because we’re taught during the PhD, that you should be an expert in your field. But I must acknowledge the obvious, that I don’t know everything, and I will never know everything. Now, what’s more important to me is being able to say “I don’t know, but I’m going to learn it.” That’s what this notebook is about.
my notebook
I will document things that I learn, especially things I’d be embarrassed to admit three years in that I don’t fully understand. I am no longer content with surface level understanding. I’m hoping this notebook can be a research equivalent to “building in public”. These notes are primarily for myself, but I am making it public for the reasons mentioned above.
The content will be broad, ranging from insights from papers, to trying my showerthought ideas, sharing ideas that failed, and generally anything I find helped me in research. I’ll also use this to practice writing3 and communicating effectively. And most importantly, it’s a reminder to myself that the only person I need to compare myself against is my past self.