R-Coffee

R-Coffee is an RNA-aware extension of the T-Coffee multiple sequence alignment framework that incorporates secondary structure information to improve the alignment of non-coding RNA sequences. Developed by Cedric Notredame and colleagues, R-Coffee addresses a critical limitation of pure sequence-based alignment methods for RNA: RNA molecules fold into conserved secondary structures — stem-loop motifs, hairpins, pseudoknots — that are often more evolutionarily conserved than their primary sequences. Two RNA molecules can share a common structure and function while having diverged beyond sequence recognition, making structure-based alignment constraints essential for accurate comparative analysis.

The method uses predicted or experimentally determined secondary structures to generate structure-based pairwise alignments, which are then incorporated into T-Coffee's consistency library alongside sequence-based alignments. The combined library guides the progressive alignment, producing alignments that respect both sequence similarity and structural compatibility. R-Coffee has been particularly valuable for aligning non-coding RNAs such as ribosomal RNA, transfer RNA, and microRNA precursors, where structure-function relationships dominate evolutionary conservation.

R-Coffee extends the T-Coffee philosophy to RNA: when sequence conservation fails, structure conservation takes over. The method is a case study in how biological alignment tools must follow the molecule — not imposing a universal model but adapting to the specific evolutionary constraints of the sequences being aligned. For RNA, that means secondary structure. For proteins, it means tertiary structure. For DNA, it often means regulatory architecture. The general principle is that the best alignment method is the one that knows what biological signal matters most for the molecules in question.