August 3, 2024 | Cameron L.M. Gilchrist, Milot Mirdita, Martin Steinegger
FoldMason is a novel progressive multiple structural alignment (MSTA) method that leverages the structural alphabet from Foldseek, a pairwise structural aligner, to align hundreds of thousands of protein structures. It computes confidence scores, offers interactive visualizations, and provides essential speed and accuracy for large-scale protein structure analysis. FoldMason uses an all-vs.-all comparison of protein structures, followed by a minimum spanning tree to guide the progressive alignment. It supports iterative refinement of alignments to maximize the Local Distance Difference Test (LDDT) score. Benchmarks show that FoldMason produces MSTAs of comparable quality to gold-standard methods while operating several orders of magnitude faster. FoldMason is particularly effective at handling flexible proteins and can be used for phylogenetic analysis, as demonstrated by its successful reconstruction of phylogenies of viral glycoproteins. The software is freely available under the GPL-3.0 license.FoldMason is a novel progressive multiple structural alignment (MSTA) method that leverages the structural alphabet from Foldseek, a pairwise structural aligner, to align hundreds of thousands of protein structures. It computes confidence scores, offers interactive visualizations, and provides essential speed and accuracy for large-scale protein structure analysis. FoldMason uses an all-vs.-all comparison of protein structures, followed by a minimum spanning tree to guide the progressive alignment. It supports iterative refinement of alignments to maximize the Local Distance Difference Test (LDDT) score. Benchmarks show that FoldMason produces MSTAs of comparable quality to gold-standard methods while operating several orders of magnitude faster. FoldMason is particularly effective at handling flexible proteins and can be used for phylogenetic analysis, as demonstrated by its successful reconstruction of phylogenies of viral glycoproteins. The software is freely available under the GPL-3.0 license.