GOPREDSIM Embeddings Adapted to FANTASIA v2

This file, embeddings_fantasia_gopredsim_2022.dump, is a PostgreSQL custom-format dump containing a comprehensive relational database. It was generated from the massive TOPGO data provided by the goPredSim project and includes data for 309,278 protein entries along with 2,243,531 annotations, as verified against the original TOPGO file.

In addition, the database incorporates a mechanism developed by our team for generating protein embeddings using the Rostlab/prot_t5_xl_uniref50 model into a relational database supporting vectors. To utilize this dump within Fantasia, simply configure the embeddings_url parameter with the URL of this Zenodo file. Fantasia’s containerized environment fully supports this process through the initialize command, which sets up the necessary database structure and loads the dump automatically.

Restoration Instructions

To load this dump into PostgreSQL, you must use pg_restore v16 or higher. Ensure database is clean and run the following command:

pg_restore -U usuario -h localhost -p 5432 -d BioData ./embeddings_fantasia_gopredsim_2022.dump

Caution: The restoration process requires pg_restore version 16 (or higher) to ensure full compatibility with all database structures contained in this dump.