Predicting SARS-CoV-2 Spike Glycoprotein Mutations

• 

  Project poster

Inspiration

I chose this problem because of its relevance to the current situation we live in. The COVID-19 pandemic has dominated our lives, and I wanted to explore how AI could help us.

What it does

I experimented with a type of neural network called LSTM. LSTM networks are capable of learning long-term dependencies such as time-series data, and I used it to try to predict future mutations within the SARS-CoV-2 virus spike protein.

How I built it

The entire project was written in Python, and I used the Keras framework to build and train my models.

Challenges I ran into

The main challenge I faced was to process the virus sequences for the model to understand meaningfully. This has taken a lot of trial and error, and is something I continue to work on.

Accomplishments that I'm proud of

While the performance of my model needs to be improved on, I'm very happy with everything this project has taught me in applying machine learning.

What I learned

I've learned many different tools one can use when building a machine learning project. As well, I've learned many of the details behind recurrent neural networks that previously were a mystery to me.

What's next for Predicting SARS-CoV-2 Spike Glycoprotein Mutations

I will continue to improve my results and predictions, because I think this is an important project to work on. Likely, I will need to investigate the dataset I am using at a deeper level.

Built With

• keras
• python