T-cell binding prediction challenge (TCR-pMHC)

Challenge Aim

This challenge focuses on the binding between a T cell and its targets. The ability to predict this accurately has important applications in vaccine development and cancer immunotherapy. Using a dataset consisting of sequences of T-cell receptors and peptide-MHCs, together with the interaction energy predicted from structural models, participants will develop machine learning algorithms to predict the binding between a T cell and potential targets.

Team: Immuchine Learners

We won the T-cell binding prediction challenge (TCR-pMHC) by utilizing a hybrid CNN-LSTM model.

We won the Mentor's Pick Award alongwith the People's Pick Award

Please see the hackathon2021.ipynb for the code we developed.

Original instruction for this repository are Given Below

At prediction time, you are given a zip file using the --input-zip argument. For each *.npz file, output your prediction.

Data and example model

The training dataset can be found in the data/training_set.zip file, and should be extracted to the data/train folder. An example CNN model is available as a notebook and in src directory.

Example Output

You code should output a file called predictions.csv in the following format:

index,prediction
0,1
1,0
2,0

Benchmarking System

The continuous integration script in .github/workflows/ci.yml will automatically build the Dockerfile on every commit to the main branch. This docker image will be published as your hackathon submission to https://biolib.com/<YourTeam>/<TeamName>. For this to work, make sure you set the BIOLIB_TOKEN and BIOLIB_PROJECT_URI accordingly as repository secrets.

To read more about the benchmarking system click here.

biohackathon-immuchine-learners/tcr-pmhc