Evolutionary: Precomputed Multiple Sequence Alignment Cache Files

Evolutionary sequence-alignment data is provided in a precomputed archive format for practical hosting and download. The dataset contains 134,898 files within the msa_cache/ directory, indexed by a metadata.csv file. It was created by LiteFold and last updated on May 25, 2026.

Use Cases

• Train protein folding models based on precomputed evolutionary sequence alignments.
• Benchmark MSA generation and search algorithms using the provided cache files.
• Accelerate bioinformatics workflows by eliminating the need to recompute alignments from raw sequence databases.

Strengths

• Contains 134,898 precomputed files, providing a substantial resource for MSA-based workflows.
• Includes a searchable metadata.csv index for locating specific files within the archive.
• Archived format is designed for practical hosting and downloading from the Hugging Face Hub.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

LiteFold

Collection Method

Precomputed from unspecified sequence databases.

Freshness

Last updated 2026-05-25 07:04:40; freshness should be verified.