ESMAtlas: Predicted Metagenomic Protein Structures from ESMFold

ESMAtlas provides more than 700 million predicted protein structures generated by Meta AI's ESMFold language model. The dataset mirrors the ESM Metagenomic Atlas, built from protein sequences in the MGnify database. The v2023_02 release corresponds to the MGnify 2023_02 snapshot and was created using the esm.pretrained.esmfold_v1() model.

Use Cases

• Large-scale protein structure retrieval based on predicted 3D coordinates.
• Structure-based search and similarity analysis using the predicted metagenomic protein atlas.
• Benchmarking protein structure prediction models against the ESMFold-generated predictions.
• Exploring protein fold space across diverse metagenomic samples based on the MGnify source.

Strengths

• Contains more than 700 million predicted protein structures, indicating a large scale.
• Generated using Meta AI's ESMFold protein language model, a state-of-the-art method.
• Based on protein sequences from the MGnify database, a recognized metagenomic resource.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data consists of computational predictions, not experimentally validated structures.

Provenance

Source

MGnify protein sequences via the ESM Metagenomic Atlas.

Collection Method

Computational predictions generated with Meta AI's ESMFold protein language model (esm.pretrained.esmfold_v1()).

Time Range

Corresponds to MGnify snapshot 2023_02.

Freshness

Last updated 2026-05-27 13:09:49; freshness should be verified.