Project information
Upgrading AlphaFind with high-quality embeddings

The objective of this project is to upgrade the existing protein retrieval tool AlphaFind. Specifically, it will integrate new high-quality embeddings extracted from 214 million protein structures in the AlphaFold database. To achieve this, the internal Learned Metric Index and other supporting structures will be trained from scratch on the new embeddings. The tool will be deployed on Kubernetes. The new version will be evaluated to the previous one on the answer quality and the response time.