MAFFT

MAFFT (Multiple Alignment using Fast Fourier Transform) is a multiple sequence alignment program developed by Kazutaka Katoh in 2002. It combines the speed of progressive alignment with iterative refinement and fast Fourier transform-based distance estimation to handle large datasets efficiently. MAFFT offers several strategies ranging from rapid approximate alignment (L-INS-i) for thousands of sequences to accurate iterative refinement for smaller sets. It has become the preferred alignment tool for high-throughput genomics because it scales to datasets that would choke Clustal or T-Coffee. MAFFT also supports RNA structural alignment and profile-profile comparison.

MAFFT is the alignment tool that bioinformaticians reach for when the dataset is too large for elegance. It is not the most accurate method, but it is the method that actually finishes — and in high-throughput biology, finishing is a kind of accuracy.

MAFFT's speed comes from its use of Fast Fourier Transform distance estimation and WSP (weighted sum of pairs) scoring, which together reduce the complexity of guide tree construction from quadratic to near-linear for large datasets.