Skip to main content
HHalign performs pairwise alignment between two multiple sequence alignments or HMMs using HMM-HMM comparison. It’s ideal for detailed structural comparison, domain alignment, and when you need precise control over the alignment of two specific sequences or profiles.

​
Overview

HHalign takes two inputs (query and template) and produces a high-quality pairwise alignment using the HMM-HMM alignment algorithm with Maximum Accuracy (MAC) alignment.

​
Key Features

  • Pairwise HMM-HMM alignment: Align two profiles with maximum sensitivity
  • Self-comparison: Find internal repeats by comparing a sequence to itself
  • Flexible input: Accepts sequences, MSAs, or pre-built HMMs
  • MAC algorithm: Maximum Accuracy alignment for optimal results

​
When to Use HHalign

Use HHalign when you need to:
  • Compare two specific proteins: Align two sequences or profiles precisely
  • Find internal repeats: Compare a protein to itself to detect repeats
  • Align protein domains: Create detailed alignments between protein domains
  • Merge two MSAs: Combine two multiple sequence alignments
  • Validate remote homology: Generate detailed alignment for suspicious hits
For database searches, use hhblits or hhsearch instead.

​
Basic Usage

1

Align two sequences

Basic pairwise alignment between query and template:
hhalign -i query.a3m -t template.hhm -o result.hhr
2

Self-comparison for repeats

Omit the template to find internal repeats:
hhalign -i protein.a3m -o repeats.hhr
This finds the best off-diagonal alignments.
3

Output alignment in different formats

Generate alignment in A3M or FASTA format:
hhalign -i query.a3m -t template.a3m -o result.hhr -oa3m aligned.a3m

​
Common Use Cases

​
Pairwise Protein Alignment

Align two proteins with their MSAs for maximum accuracy: 
hhalign -i query.a3m -t template.a3m -o result.hhr
From the README.md example: 
hhalign -i T0187.a3m -t d1hz4a_.hhm -o result.hhr

​
Finding Internal Repeats

Detect repeating domains within a single protein: 
hhalign -i protein.a3m -o repeats.hhr -alt 3
This finds up to 3 non-overlapping alternative alignments.

​
Domain Alignment with Position Exclusion

Align specific regions while excluding others: 
hhalign -i query.a3m -t template.a3m -o result.hhr \
  -excl 1-33,97-168          # Exclude query positions
  -template_excl 1-50        # Exclude template positions

​
Generate Multiple Output Formats

Create alignment in multiple formats simultaneously: 
hhalign -i query.a3m -t template.a3m \
  -o result.hhr \
  -oa3m alignment.a3m \
  -Ofas alignment.fas

​
Key Parameters

  • -i <file> - Query alignment (A2M, A3M, FASTA) or HMM
  • -t <file> - Template alignment or HMM (omit for self-comparison)
  • -o <file> - Output file in HHR format (default: <infile>.hhr)
  • -oa3m <file> - Write alignment in A3M format
  • -aa3m <file> - Append alignment to existing A3M file
  • -Ofas/-Oa2m/-Oa3m <file> - Pairwise alignment in FASTA/A2M/A3M
  • -glob - Use global alignment (align full sequences)
  • -loc - Use local alignment (default, best local match)
  • -mact [0,1] - MAC realignment threshold (default: 0.35)
  • -norealign - Disable MAC realignment
  • -alt <int> - Number of alternative alignments to show (default: 1)
  • -smin <float> - Minimum raw score for alternative alignments
  • -excl <range> - Exclude query positions (e.g., ‘1-33,97-168’)
  • -template_excl <range> - Exclude template positions
  • -id [0,100] - Maximum pairwise sequence identity % (default: 90)
  • -cov [0,100] - Minimum coverage with master sequence % (default: 0)
  • -qid [0,100] - Minimum sequence identity with query % (default: 0)
  • -diff [0,inf] - Filter for diversity (default: depends on mode)

​
Output Format

Query         query_protein
Match_columns 250
Template      template_protein
Match_columns 245

Probabilities and scores:
 Probability: 99.8%
 E-value:     2.3e-25
 Score:       185.2 bits
 Aligned_cols: 220
 Identities:  35%
 Similarity:  0.521

Q query_protein   1 MSTIKGYRILLSGAIDSFSLTES--DKPTYLLVGPSGCSGKTTLLNAIAG  48
Q Consensus       1 ~~~i~~~~~~~sG~iD~f~l~~~--~k~~Y~l~gp~G~~g~~~ll~~~~~  48
                    |+++++++++++++++++++++++  +|+||++++++++||||||||||++
T Consensus       1 ~~~v~~~~~~~sG~iD~f~l~~~ggek~~Y~l~gp~G~~g~~~ll~~~~~  50
T template_prot   1 MRAVKGYRILLSGAIDSFSLTEGGGEKATYRLVGPSGCSGKTTLLNAIAG  50

​
Tips and Best Practices

Self-comparison tips: When finding internal repeats, use -alt to find multiple non-overlapping alignments and -smin to set a minimum score threshold.
Global alignment (-glob) forces full-length alignment. Use local alignment (default) unless you’re certain both sequences should align end-to-end.
MAC threshold: Lower -mact values (closer to 0) favor global alignment at the ends, while higher values (>0.1) favor local alignment. The default 0.35 works well for most cases.

​
Advanced Options

​
Fine-tune Alignment Quality

Control the alignment algorithm parameters: 
hhalign -i query.a3m -t template.a3m -o result.hhr \
  -mact 0.5 \      # Adjust MAC threshold
  -shift -0.2 \    # Profile-profile score offset
  -corr 0.1        # Weight for pair correlations

​
Secondary Structure Scoring

Include secondary structure in alignment: 
hhalign -i query.a3m -t template.a3m -o result.hhr \
  -ssm 2 \    # SS scoring mode (0=off, 1,2=after/during)
  -ssw 0.5 \  # SS score weight
  -ssa 0.3    # SS confusion matrix parameter

​
Custom Gap Penalties

Adjust gap opening and extension penalties: 
hhalign -i query.a3m -t template.a3m -o result.hhr \
  -gapb 1.0 \   # Transition pseudocount admixture
  -gapd 0.15 \  # Gap open admixture for deletes
  -gape 1.0 \   # Gap extend admixture
  -gapf 0.6 \   # Factor for delete gap open
  -gapg 0.6     # Factor for insert gap open

​
Multiple Alternative Alignments

Find several alternative alignments: 
hhalign -i protein.a3m -o repeats.hhr \
  -alt 5 \      # Show up to 5 alternatives
  -smin 50.0    # Minimum score threshold

​
Self-Comparison Mode

When only -i is provided (no -t), HHalign compares the query to itself: 
hhalign -i tandem_repeat.a3m -o self_alignment.hhr -alt 3
This mode:
  • Finds the best off-diagonal alignment (skips the trivial diagonal)
  • Reports all non-overlapping alignments above the significance threshold
  • Useful for detecting tandem repeats, domain duplications, or internal symmetry

​
Comparison with Other Tools

FeatureHHalignHHsearchHHblits
Input1-2 sequences/MSAs1 query + database1 query + database
OutputSingle alignmentRanked listRanked list + MSA
IterationNoNoYes
Best forPairwise comparisonStructure searchDeep MSA building
Self-comparisonYesNoNo
  • hhblits - Iterative database searching
  • hhsearch - Database searching without iteration
  • hhmake - Convert alignments to HMM format
  • hhfilter - Filter alignments

​
References

Remmert M, Biegert A, Hauser A, Söding J (2012). HHblits: lightning-fast iterative protein sequence searching by HMM-HMM alignment. Nature Methods, 9:173-175. doi: 10.1038/nmeth.1818

Build docs developers (and LLMs) love

Get started for freeTalk to us