Thal-Kak Multiple Sequence Alignment Databases
bioinformatics biomolecular modeling life sciences metagenomics msa protein protein folding structural biology
Description
A collection of prebuilt protein sequence databases for multiple sequence alignment (MSA) search. MSAs are the main input of protein structure prediction and are also used to study conservation, coevolution and protein function. The nine databases, in the native formats of MMseqs2 and HH-suite, derive from UniProt UniRef100 (release 2026_01), the reduced BFD, MGnify (release 2024_04), EnVhogDB and the Logan protein catalogue from the NCBI Sequence Read Archive. The sequences of UniRef100, MGnify and the reduced BFD were re-clustered, EnVhogDB was re-clustered at its upstream standard level, and Logan was built into an MMseqs2 expandable-profile database. Each database is published as one .tar.zst archive. They are set up for Thal-Kak, a pipeline that unifies several structure prediction models, so that its local MSA search writes MSAs in the input format of those models.
Update Frequency
Irregularly. New releases are added as new archives.
License
Each database carries the license of the source it derives from; no database mixes sources. LICENSE.txt at the bucket root is the full record of upstream licenses, attribution, citation and changes.
| Archives | License | Source |
|---|---|---|
| hhblits/uniref100_2026_01 | CC BY 4.0 | UniProt UniRef100, release 2026_01 |
| mmseqs/bfd_reduced | CC BY-SA 4.0 | BFD reduced subset, via the AlphaFold databases |
| mmseqs/mgnify_clusters, hhblits/mgnify_clusters | CC0 1.0 | MGnify protein database, release 2024_04 |
| mmseqs/envhog_std, hhblits/envhog | CC BY 4.0 | EnVhogDB |
| mmseqs/logan_human, hhblits/logan_human, hhblits/logan_nonhuman | No license is granted. NCBI Policy and NIH Genomic Data Sharing Policy | Logan v1 (SRA cutoff 2023-12-10) |
Documentation
https://github.com/CSSB-SNU/Thal-Kak/tree/main/aws_open_data
Managed By
CSSB, Seoul National University
See all datasets managed by CSSB, Seoul National University.
Contact
https://github.com/CSSB-SNU/Thal-Kak/issues
How to Cite
Thal-Kak Multiple Sequence Alignment Databases was accessed on DATE from https://registry.opendata.aws/thal-kak-msa. For the pipeline that searches these databases, cite Bae J. et al. Thal-Kak: unifying biomolecular structure predictors reveals a sampling-selection gap. bioRxiv 2026.08.19.745680 (2026). doi:10.64898/2026.08.19.745680. Also cite the upstream database of each archive you use, as given in LICENSE.txt; for the CC BY and CC BY-SA databases this attribution is a license condition.
Usage Examples
Tutorials
-
Get To Know A Dataset: Thal-Kak Multiple Sequence Alignment Databases by Kwanwook Kim (CSSB, Seoul National University)
Tools & Applications
Publications
-
Thal-Kak: unifying biomolecular structure predictors reveals a sampling-selection gap by Bae J., Jo S., Kim Y., Kim D., Kim K., Park S., et al.
Resources on AWS
-
- Description
- MMseqs2 expandable profile databases, one .tar.zst archive per database (mgnify_clusters, envhog_std, bfd_reduced, logan_human). Each archive holds four MMseqs2 databases; build the .idx search index after extraction.
- Resource type
- S3 Bucket
- Amazon Resource Name (ARN)
arn:aws:s3:::thal-kak-msa/mmseqs/- AWS Region
us-east-1- AWS CLI Access (No AWS account required)
aws s3 ls --no-sign-request s3://thal-kak-msa/mmseqs/
-
- Description
- HH-suite (FFindex) databases, one .tar.zst archive per database (uniref100_2026_01, mgnify_clusters, logan_nonhuman, envhog, logan_human). Each archive holds the a3m, hhm and cs219 databases.
- Resource type
- S3 Bucket
- Amazon Resource Name (ARN)
arn:aws:s3:::thal-kak-msa/hhblits/- AWS Region
us-east-1- AWS CLI Access (No AWS account required)
aws s3 ls --no-sign-request s3://thal-kak-msa/hhblits/
-
- Description
- Notifications for new Thal-Kak Multiple Sequence Alignment Databases data
- Resource type
- SNS Topic
- Amazon Resource Name (ARN)
arn:aws:sns:us-east-1:029678959616:thal-kak-msa-object_created- AWS Region
us-east-1