My practicum site was Lieber Institute for Brain Development, a private research institute in Baltimore, Maryland. The institute focuses on making clinical advances in psychiatric brain disorders such as schizophrenia. Unfortunately, I could not be physically at the site due to covid restrictions, so my project was remote. Two years ago I did an internship at Lieber Institute working on data visualization and R programming package development, so this site was my number one pick for the project. I found my site supervisor very easily since he is my father Richard Straub, a Senior Investigator at the institute. My advice to any future SGCer is to think about what possible practicum sites are in your backyard, ie. people that you already know, friends, family, etc. that may have cool projects for you to work on.
During this project, I was tasked with a wide variety of duties. My first task was to build and set up the computer that I used for the project. This task involved physically building the computer, installing the CPU, GPU, ram, etc., and also downloading and installing all of the necessary software for the project such as Anaconda Navigator, VScode, R studio, and Nvidia cuda drivers. My next task was to research what possible machine learning algorithms I could apply to the dataset provided. I needed to find algorithms that used classification since we are trying to predict Schizophrenia vs non-Schizophrenia, and ones that worked with how few observations our data had. Next, I spent time learning how to do proper exploratory data analysis so I could use it to subset the data in meaningful ways. I then wrote code in both R and Python setting up Random Forest, Linear Classifier, and Gradient Boosted Trees so I could run the subsets I had created through. My next task was to pull the meaningful information from the models (algorithm runs) such as the variable importance lists and confusion matrices. I then analyzed what genes were important in prediction for each algorithm and found commonalities in what genes were important in multiple different algorithms to present to my site supervisor.
In this project I worked with gene expression and electrophysiological data. Throughout, I learned what electrophysiology was, how it related to brain activity between schizophrenics and controls, and how we could use the data to help classify which genes are important in predicting schizophrenia. I also learned about how genes are expressed and how that relates to protein production, which is often different between schizophrenics and controls.
Working on this practicum project was an eye-opening experience into how projects are completed in real-world research. During my time working on this project I found the connection between scientific research and actual people to be quite fascinating. The data that I was working with contained measurements from real people’s brains and genomes (both people with and without schizophrenia), and using these measurements we can try to figure out methods of helping out people with the same disease. In this case, there is a direct connection between societal mental health issues and the science I was working with, and I have gained a far greater appreciation for that connection during this project. This project has also pushed me to explore taking more biology classes in the future. Combining biology and computer science in this project was extremely interesting to me since it allowed me to apply my computer science to gain insight into something that could possibly help people. Working on this project has confirmed what I had originally thought about my future career plans, Biotech may be a good option. The intersection between biology and computer science will only continue to expand as new technology comes out, and it will be very interesting to see what the next big revolution in Biotech will bring.
Back To Main Site: