FunClust

FunClust identifies structural motifs in non-homologous protein structures by detecting spatially conserved local substructures that may indicate shared function.

Key Features:

• Unsupervised Identification: Operates in an unsupervised mode accepting sets of protein structures without prior motif knowledge.
• Non-Contiguous Motif Detection: Detects structural motifs composed of as few as three residues that need not be contiguous or in the same sequence order.
• Sequence Independence: Focuses on spatial conservation independent of amino acid sequence to reveal structural similarities.
• Local Structural Comparison: Uses a local structural comparison algorithm to identify subsets of similar amino acids between each pair of input protein chains.
• Clustering of Matches: Applies a clustering procedure to group pairwise similarities across different structure pairs to highlight shared motifs.

Scientific Applications:

• Functional Inference from Structure: Infers potential common functions among proteins that lack detectable sequence similarity by identifying conserved structural motifs.
• Identification of Binding Sites and Epitopes: Reveals shared structural features that can correlate with ligand-binding sites or common epitopes across non-homologous proteins.
• Comparative Structural Analysis: Identifies largest common structural motifs across test sets of non-homologous proteins and correlates them with known functional features.

Methodology:

Performs pairwise local structural comparisons to identify spatially conserved amino acids between protein chains, then clusters the identified similarities across structures and reports motifs present in at least three different protein structures.