QSAR Analysis with Galileo

Inspiration

This project was inspired by my interest in learning data science, goal of doing the work myself instead of delegating to someone with more data science experience, and Massachusetts going into phase 3 for COVID-19 vaccination. Because data science is inherently interdisciplinary, I was curious about how far I'd get exploring a database that bio/pharmaceutical R&D professionals would review.

What it does

QSAR stands for quantitative structure–activity relationship and QSAR models show supposed relationships between chemical structures and biological activity. Studying protein biological activity is relevant for exploring disease pathology or related biomarkers.

I explored and performed some statistical test on the bioactivity of some of the compounds known to inhibit the protein of coronavirus 3c-like proteinase, which is essential for coronavirus replication.

How we built it

I used Galileo's platform.

Challenges we ran into

There were a lot of unknowns - how do drugs work, what info can be found, what can I learn in a weekend, will I really understand what's being done. And I took it in stride, one step at a time — as the saying goes, a journey of a thousand miles begins with a single step.

Accomplishments that I'm proud of

This is my first solo data science project, and I learned a lot about different domains and see the importance of collaborating with subject matter experts (SMEs).

What I learned

I learned more about how drug discovery works, the biology of proteins, how the chemical structure of proteins affect solubility, and some useful transformations and tests for statistics. I also learned that protein-based vaccines are only one of many types of drugs that are out there, so this is a sliver of what's out there. Much respect for all the researchers — and frontline workers — out there during these unprecedented times!

What's next for QSAR Analysis with Galileo

Using data to generate ML model to predict molecule solubility based on structure.

Built With

• colab
• intel-galileo
• jupyter
• python