Alignment and phylogenetics · Software
MAFFT
A multiple sequence alignment program for amino acid or nucleotide sequences.
What MAFFT is
MAFFT is a multiple sequence alignment tool designed for amino acid and nucleotide sequences. Originally developed by Kazutaka Katoh and colleagues, it provides researchers with algorithms tailored to different dataset sizes and alignment challenges, categorised under alignment and phylogenetics.
The software addresses the computational complexity of generating multiple sequence alignments across large numbers of sequences or long sequences. It allows users to choose between high-accuracy progressive and iterative methods for small sequence counts and high-speed heuristic methods suitable for tens of thousands of sequences.
MAFFT supports specialized tasks such as full-length multiple sequence alignments of closely related viral genomes, non-coding RNA alignments that integrate structural details, profile alignments, and the addition of unaligned sequences or fragments to existing reference alignments.
How it works
MAFFT employs strategies ranging from fast progressive alignments relying on Fast Fourier Transform methods (such as FFT-NS-1 and FFT-NS-2) to iterative refinement strategies (such as FFT-NS-i, L-INS-i, G-INS-i, and E-INS-i). For very large collections of unaligned sequences, it can employ the PartTree algorithm to construct an approximate guide tree rapidly.
The program accepts sequences in FASTA format and automatically recognizes whether the input represents proteins or nucleic acids. Depending on the command or flags chosen (such as `--auto`, `mafft-linsi`, or default `mafft`), it balances accuracy against speed. It also includes methods to parallelise calculations using multithreading and MPI, as well as methods to incorporate RNA structure or limit over-alignment.
Key facts
- Latest version
- 7.526
- Input format
- FASTA format
- Supported sequence types
- Amino acid or nucleotide sequences
- High-accuracy mode
- L-INS-i (typically for < ~200 sequences x ~2,000 sites)
- High-speed mode
- FFT-NS-2 (handles < ~30,000 sequences)
- Supported operating systems
- Unix-like systems (Linux, Mac OS X) and Windows
What you can use it for
- High-accuracy protein sequence alignment
- Align small datasets (fewer than approximately 200 sequences of up to roughly 2,000 sites) using the iterative L-INS-i algorithm.
- Large-scale multiple sequence alignment
- Quickly align large datasets containing up to roughly 30,000 sequences using the FFT-NS-2 algorithm or approximate trees generated by PartTree.
- Adding sequences to existing alignments
- Incorporate newly discovered unaligned sequences or sequence fragments into a pre-computed alignment without re-aligning the entire dataset, using the --add and --addfragments options.
- Non-coding RNA structural alignment
- Align non-coding RNAs by incorporating secondary structural information directly into the alignment workflow.
- Viral genome alignment
- Perform full-length multiple sequence alignments on closely related viral genomes.
Run it here
MAFFT is preinstalled on the machine behind every research conversation. Ask the agent in plain language, or give it the exact command — it runs it, reads the output and explains it.
Strengths and limits
Good at
- Offers scalable methods ranging from fast Fourier transform heuristics for large inputs to accurate iterative algorithms for smaller inputs.
- Automatically identifies whether sequences are amino acids or nucleotides from the input FASTA file.
- Includes an automatic parameter selection flag (--auto) when optimal strategy settings are uncertain.
- Provides functionality to add new sequences or fragments to existing profile alignments without realigning from scratch.
- Supports computational acceleration via multithreading and MPI parallelisation.
Watch out for
- The accurate L-INS-i method is restricted to smaller datasets of fewer than approximately 200 sequences and 2,000 sites.
- Versions 7.463 to 7.486 contained a memory allocation bug in FFT-NS-i and --auto that caused calculation failures, requiring version 7.487 or higher.
Questions
Can I run MAFFT online without installing it?
MAFFT is preinstalled on the machine behind every research conversation. Ask the agent in plain language, or give it the exact command — it runs it, reads the output and explains it.
What licence is MAFFT under?
MAFFT is published under its authors' own licence; the project page at https://mafft.cbrc.jp/alignment/software/ states it.
What input format does MAFFT require?
MAFFT requires sequences in FASTA format. It automatically determines whether the sequences consist of nucleotides or amino acids.
Which option should be used when the ideal alignment strategy is unknown?
Users can run MAFFT with the `--auto` argument, which allows the program to select an appropriate alignment strategy based on the data size.
Can MAFFT align non-coding RNAs?
Yes, MAFFT includes frameworks that incorporate RNA structural information for aligning non-coding RNAs.
How can new sequences be merged into a reference alignment?
MAFFT provides the `--add` and `--addfragments` options to integrate new, unaligned sequences or fragments directly into an existing alignment.
How to cite it
Katoh, Standley 2013 ( Molecular Biology and Evolution 30 :772-780 ) MAFFT multiple sequence alignment software version 7: improvements in performance and usability.
Written from MAFFT's own README and project page. Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.