Derivation table. The table is made of one cycle per row and their corresponding nucleotides in the columns. The colors match the colors in Figure 1.

Simplified example of the closest derived sarcin–ricin sequences in an alignment. (A) Two derived sarcin–ricin motif sequences. (B) All possible matches (bold underlined) of the two sequences in the alignment. (C) One structural site in the reference (E. coli) sequence. The first strand matches in the reference sequence at position 20, and the second strand matches at position 40. For each sequence of the alignment, we choose the positions (bold underlined), among all possible matches, that minimizes the Manhattan distance. (D) Resulting alignment. The closest matches, in each sequence, are shown in bold-underlined characters.

The sarcin–ricin motif. (A) Stereoview of the 3D structure. The nucleotides are labeled by the X_i (5′-strand) and Y_i (3′-strand). The backbone is shown using a light green cylinder. Nitrogen atoms are in blue; oxygen in red; and carbon in green. The hydrogen atoms are not shown. (B) Tertiary structure and cycles. A minimal cycle basis of the sarcin–ricin motif is made of five minimum cycles: C₁ to C₅. The symbols used to indicate base stacking and base pairing are described in the Materials and methods section (see arc-type nomenclature). The backbone interactions are shown with bold lines. (C) The same minimum cycle basis as in (B), but without the backbone interactions (indicated by dotted lines).

Graph-grammar derivation (on a reduced set of sequences). The box on the left indicates the set of sequences found for each cycle. The derivation table is on the right. (A) Insertion of the first C₁ sequence: GUAA (in red). (B) Insertion of the first C₃ sequence, AGU (in green). This insertion is not possible because the first nucleotide of C₃, A, does not match with the first nucleotide of C₁, G, which was previously inserted in the table. Since no other sequence is available for C₃, the algorithm backtracks to the previous cycle, C₁, and selects its next sequence, AUAA. (C) Last step. The insertion of the C₅ sequence, GAA, completes the sequence of the entire motif and represents a valid derivation of the graph-grammar: AGUA/GAA. The order of the nucleotides corresponds to the order given by the labels in Figure 2.

Hypothetical sarcin–ricin structure. (A) Tertiary structure and cycles. The shortest cycle basis of the hypothetical sarcin–ricin structure shows five cycles, C₁ to C₅, characterized by canonical Watson–Crick base pairs. The backbone interactions are shown with bold lines. (B) Stereoview of the MC-Fold/MC-Sym model (2.1 Å RMSD). The model (colored) is superimposed on the seed motif (green). The nucleotides are labeled by the X_i (5′-strand) and Y_i (3′-strand). The backbone of the seed motif is shown using a light green cylinder. The backbone of the model is not shown. The nitrogen atoms in the model are in blue; oxygen in red; and carbon in gray. The carbon atoms shown in yellow emphasize the unconventional inward stacking between X₂ and Y₁, a characteristic feature of the sarcin–ricin motif. X₂ and X₃ do not pair. The hydrogen atoms are not shown. (C) Stereoview of the alignment model (0.9 Å RMSD). The model (colored) is superimposed on the seed motif (green). The color and numbering nomenclature is the same as in (B). X₂ and X₃, U and U, base pair as in the seed motif.

Alignment of the bacterial 23S rRNA sequences. The alignment contains 806 sequences. The alignment was broken in five alignments: (A–E). Each smaller alignment was made of the sequences that were not derived by the graph-grammar and their above and below sequences. The parentheses represent canonical base pairs. The braces and brackets represent non-canonical base pairs. The tilde characters, ‘∼’, are used for the unpaired nucleotides. Each sarcin–ricin site is assigned a loop number (i.e. L11, L13, etc.). The last site is span by too many nucleotides to be assigned a loop. Each sequence is given a unique number (on the right side). The source species of the sequences are indicated on the left. The sequence of E. coli is the structural reference, indicated by ‘#’ character. The nucleotides shown in bold and underlined correspond to the sarcin–ricin sites that are derived by the graph-grammar. (A) Site ‘B’204-‘B’205-‘B’206-‘B’207/‘B’189-‘B’190-‘B’191 in chain ‘B’ of PDB entry 2AWB, corresponding to position (63-48) in the alignment. (B) ‘B’241-‘B’242-‘B’243-‘B’244/‘B’254-‘B’255-‘B’256; (28–42). (C) ‘B’371-‘B’372-‘B’373-‘B’374/‘B’400-‘B’401-‘B’402; (12–43). (D) ‘B’457-‘B’458-‘B’459-‘B’460/‘B’469-‘B’470-‘B’471; (25–38). (E) ‘B’1265-‘B’1266-‘B’1267-‘B’1268/‘B’2012-‘B’2013-‘B’2014; (5–64).