Foldseek

Foldseek performs rapid, sensitive comparisons of large protein structure sets to enable fast structural similarity searches for evolutionary and functional inference.

Key Features:

• Structural alphabet representation: Represents tertiary interactions within proteins as sequences over a structural alphabet, converting 3D structure into sequence-like data.
• Sequence-based alignment: Aligns query protein structural sequences against large structure databases to identify structural matches.
• High speed: Achieves computational efficiencies four to five orders of magnitude greater than traditional structural aligners.
• Maintained sensitivity: Maintains high sensitivity, achieving 86%, 88%, and 133% of the sensitivities of Dali, TM-align, and CE, respectively.

Scientific Applications:

• Evolutionary inference: Rapidly compares structures to explore evolutionary relationships among proteins.
• Functional annotation: Identifies functional similarities between proteins based on structural matches.
• Motif discovery: Detects novel structural motifs across diverse protein structure datasets.

Methodology:

Represents tertiary interactions as sequences over a structural alphabet and performs sequence-based alignments of those structural sequences against large structure databases.