We won the Kaggle competition "Open Problems - Single-Cell Perturbations"!

🥈 I’m thrilled to announce that Martin Garrido Rodriguez-Cordoba and I have won the Judges’ Award in the “Single-Cell Perturbations” Kaggle challenge, as part of the NeurIPS 2023 Competition Track. Our Machine Learning model also earned a Silver Medal in the public leaderboard, achieving a top <2% performance by ranking 16th out of 1,097 teams and securing the 3rd position on public leaderboard.

Neural network architecture used for the challenge (ScAPE model)

The challenge was centered around using Machine Learning to predict how human peripheral blood cells of healthy donors respond to drugs at the gene expression level. By using single-cell RNA-seq data, this competition aimed to bridge computational biology and machine learning, enhancing our capabilities in predicting drug effects on diverse cell types and conditions.

I would like to thank the Open Problems in Single-Cell Analysis organization, Cellarity, and Kaggle for organizing this fascinating competition, as well as all the competitors who contributed to its success.

If you are interested in our solution or want to learn more about our approach to this challenge, please check our detailed post on Kaggle. We’ve also released the code on GitHub for those who want to dive deeper into our work.