Summer Internship at Pardee Lab

Throughout this summer, I’ve been working at Pardee Lab at University of Toronto. This was absolutely a precious yet challending experience.

Group Photo

Here is the group photo of the lab, Doctor Keith, my supervisor, is at the fourth counting from the left of the first row, while I’m at the fifth counting from the left of the second row.

Brief Introduction of My Task

Although Pardee Lab is a bio-focused lab, as a computer science undergrad, I participated in its computational project. The project aims at developing machine-learning guided biosimilar protein structure prediction combining with web-lab verification to really find some biosimilar proteins that may be easily to produce. In one sentence, we aim at using machine learing to increase the efficiency of bio-research. (what our team has accomplished in summer is marked as red in flowchart)

My task is about setting up a steady environment for alphafold2 - a machine learning software that can predict 3D-structure of a given amino sequence based on deep-learning. The problem is that it consumes extreme amount of resources like storage space and GPU.

Obviously, it’s hard to setup a physical server in lab in short time, so I put my sight on google cloud server - a flexible and affordable alternative.

I overcame technical difficulties in alphafold setup and developed (automation pipeline)[https://github.com/zhuyuezx/Pardee-Lab-computation-project-pipeline] to helpe those non-professional users. At the end of the summer I also made a tutorial about the whole setup process:

It is reassuring that I’ve really used what I’ve learned from CS courses, like writing shell script for automation and linux commands to remote connect and operate on cloud servers. Moreover, lab-style weekly meeting and seminar really brought me the experience about how a lab operates and how I should conform with its pace to make my own contributions.

Thoughts After Internship

Although being a CS student and it’s hard for me to understand those professional biology terms, I can use my skills to genuinely help to boost the efficiency of bio research. Automation pipeline, cloud server setup, and software-assisted data analysis, these are far less than I can bring to my lab.

Moreover, our team has students major in bio-informatics, statistics, computer science, chemistry, and engineering. As long as we can communicate effectively, the power of interdisciplinary cooperation helped us to come up with novel ideas and discover new research directions.

I believe that interdisciplinary research would become more and more common in the furture while computer science must be a indispensible part of it.

Modified Green Fluorescence Protein(GFPP) illustrated in 3D