---
license: mit
---
This is a dataset of 10,000 interacting pairs of proteins obtained from UniProt, 
and clustered using methods explain in [this blog post](https://huggingface.co/blog/AmelieSchreiber/faster-pha). Note, 
cluster 0 is over-represented in this dataset and this should be considered when creating train/test splits with this
data.