3D-Coffee

3D-Coffee is a structure-aware extension of the T-Coffee multiple sequence alignment method that incorporates three-dimensional protein structural information to guide sequence alignment. Developed by Cedric Notredame and colleagues, it addresses a fundamental limitation of pure sequence-based alignment methods: when protein sequences have diverged beyond the twilight zone of sequence similarity — typically below 25-30% identity — sequence conservation alone is insufficient to identify homologous positions. Structural superposition of known 3D structures provides a gold standard that transcends sequence divergence, and 3D-Coffee exploits this by using structural alignments as additional constraints in its consistency-based framework.

The method operates by retrieving known structures for sequences in the input set from databases such as PDB (Protein Data Bank), then generating both sequence-based pairwise alignments and structure-based alignments for all pairs with structural information. These structural alignments are weighted more heavily than sequence alignments in the consistency library, effectively telling the algorithm: trust