Skip to main content

​
Overview

HH-suite can search against various pre-built protein databases optimized for remote homology detection. These databases range from comprehensive sequence collections to specialized structure and domain databases.

Uniclust30

Comprehensive clustered protein sequences at 30% identity

BFD

Big Fantastic Database with 2.5 billion environmental sequences

PDB70

Representative protein structures from the PDB

Pfam

Curated protein family database

​
Uniclust30

​
Description

Uniclust30 is a comprehensive protein sequence database clustered at 30% sequence identity. It provides excellent coverage for homology detection while maintaining reasonable database size. Key Features:
  • Clustered from UniProt at 30% sequence identity
  • Updated regularly
  • Optimized for HHblits iterative searches
  • Good balance of sensitivity and speed

​
Download

wget http://wwwuser.gwdg.de/~compbiol/uniclust/2023_02/UniRef30_2023_02_hhsuite.tar.gz
tar xzvf UniRef30_2023_02_hhsuite.tar.gz

​
Usage

hhblits -i query.fasta -d UniRef30_2023_02 -o results.hhr -oa3m query.a3m

​
Reference

Mirdita M, von den Driesch L, Galiez C, Martin MJ, Söding J, Steinegger M (2017) Uniclust databases of clustered and deeply annotated protein sequences and alignments. Nucleic Acids Research, 45(D1):D170-D176. doi: 10.1093/nar/gkw1081
Uniclust30 is the recommended database for most HHblits searches. It provides the best balance between sensitivity, speed, and database size.

​
BFD

​
Description

The Big Fantastic Database (BFD) contains approximately 2.5 billion protein sequences, mostly from environmental samples. It provides maximum sensitivity for detecting remote homologs. Key Features:
  • 2.5+ billion sequences
  • Mostly environmental (metagenomic) sequences
  • Highest sensitivity for remote homology detection
  • Significantly larger and slower than Uniclust30
  • Used by AlphaFold for MSA generation

​
Download

wget https://bfd.mmseqs.com/bfd_metaclust_clu_complete_id30_c90_final_seq.sorted_opt.tar.gz
tar xzvf bfd_metaclust_clu_complete_id30_c90_final_seq.sorted_opt.tar.gz
BFD is extremely large (>1 TB uncompressed). Ensure you have sufficient disk space and memory before downloading.

​
Usage

hhblits -i query.fasta -d bfd_metaclust_clu_complete_id30_c90_final_seq \
  -o results.hhr -oa3m query.a3m -n 3 -cpu 8

​
Reference

Steinegger M, Söding J (2019) Clustering huge protein sequence sets in linear time. Nature Communications, 10:2542. doi: 10.1038/s41592-019-0437-4

​
PDB70

​
Description

PDB70 is a filtered subset of protein structures from the Protein Data Bank, clustered at 70% maximum sequence identity. It’s ideal for structure-based searches and homology modeling. Key Features:
  • Representative protein structures from PDB
  • Clustered at 70% sequence identity
  • Includes secondary structure information
  • Updated weekly
  • Essential for structure prediction and modeling

​
Download

wget http://wwwuser.gwdg.de/~compbiol/data/hhsuite/databases/hhsuite_dbs/pdb70_from_mmcif_latest.tar.gz
tar xzvf pdb70_from_mmcif_latest.tar.gz

​
Usage

hhsearch -i query.hhm -d pdb70 -o results.hhr -atab results.atab
For structure modeling: 
hhsearch -i query.hhm -d pdb70 -o results.hhr
hhmakemodel.py -i results.hhr -ts template.pdb -o model.pdb
PDB70 searches are typically performed with hhsearch after building an HMM profile from a multiple alignment, not directly with hhblits.

​
Pfam

​
Description

Pfam is a curated database of protein families, each represented by multiple sequence alignments and HMMs. It’s useful for domain annotation and functional classification. Key Features:
  • Manually curated protein families
  • High-quality seed alignments
  • Comprehensive functional annotation
  • Domain architecture information
  • Standard for protein family classification

​
Download

wget http://wwwuser.gwdg.de/~compbiol/data/hhsuite/databases/hhsuite_dbs/pfamA_35.0.tar.gz
tar xzvf pfamA_35.0.tar.gz

​
Usage

hhsearch -i query.hhm -d pfamA_35.0 -o results.hhr

​
Reference

Mistry J, et al. (2021) Pfam: The protein families database in 2021. Nucleic Acids Research, 49(D1):D412-D419. doi: 10.1093/nar/gkaa913

​
SCOP

​
Description

Structural Classification of Proteins database, organized hierarchically by fold, superfamily, and family.

​
Download

wget http://wwwuser.gwdg.de/~compbiol/data/hhsuite/databases/hhsuite_dbs/scop70_1.75.tar.gz
tar xzvf scop70_1.75.tar.gz

​
Usage

hhsearch -i query.hhm -d scop70_1.75 -o results.hhr

​
Additional Databases

​
MPI Bioinformatics Toolkit Databases

The MPI Bioinformatics Toolkit maintains additional specialized databases:
  • COG - Clusters of Orthologous Groups
  • ECOG - Evolutionary Genealogy of Genes
  • CDD - Conserved Domain Database
  • dbCAN - Carbohydrate-Active Enzymes database
  • SMART - Simple Modular Architecture Research Tool
wget http://ftp.tuebingen.mpg.de/pub/ebio/protevo/toolkit/databases/hhsuite_dbs/

​
Database Selection Guide

For general homology searches:
  • Start with Uniclust30 - best balance of speed and sensitivity
  • Use BFD if you need maximum sensitivity and have computational resources
For structure prediction:
  • Use PDB70 to find structural templates
  • Search after generating an MSA with Uniclust30 or BFD
For domain/family annotation:
  • Use Pfam for standard family classification
  • Use SCOP for structural classification
For specialized searches:
  • Use domain-specific databases (COG, dbCAN, etc.) from MPI Toolkit

​
Database Formats

HH-suite databases consist of several files: 
database_a3m.ffdata       # A3M alignments (FFindex format)
database_a3m.ffindex      # A3M alignment index
database_hhm.ffdata       # HMM profiles (FFindex format)
database_hhm.ffindex      # HMM profile index
database_cs219.ffdata     # Context-specific profiles (optional)
database_cs219.ffindex    # CS219 index (optional)
All files with the same prefix must be present in the same directory for the database to work properly.

​
Performance Considerations

​
Database Size vs. Speed

DatabaseSequencesDisk SpaceSearch TimeSensitivity
Uniclust30~100M~100 GBFastGood
BFD~2.5B>1 TBSlowExcellent
PDB70~50K<5 GBVery FastStructure-specific
Pfam~20K<2 GBVery FastFamily-specific

​
Memory Requirements

BFD searches require significant RAM:
  • Minimum: 64 GB
  • Recommended: 128+ GB for efficient searching
  • Consider using -cpu to limit parallelization if memory-constrained

​
Building Custom Databases

You can create custom databases from your own sequences. See the Building Custom Databases guide for detailed instructions. 
# Quick example
hhsuitedb.py -ia3m my_alignments/ -o my_database

​
Database Updates

Databases are typically updated on the following schedules:
  • Uniclust30: Every 2-3 months
  • BFD: Annually
  • PDB70: Weekly (follows PDB releases)
  • Pfam: Every 6-12 months
Check the HH-suite database repository regularly for new releases.

​
Troubleshooting

​
Database Not Found

Error: Database files not found
Solution: Ensure all database files (.ffdata, .ffindex) are in the specified directory and use the correct basename without extensions.

​
Out of Memory

Error: Cannot allocate memory
Solution:
  • Reduce number of CPUs with -cpu option
  • Use a smaller database (Uniclust30 instead of BFD)
  • Increase system swap space

​
Corrupted Database

Error: Could not read database entry
Solution: Re-download and extract the database. Verify checksums if provided.

​
See Also

Build docs developers (and LLMs) love

Get started for freeTalk to us