Figure 2. Principle of triplet elements that contain both sequence and structural information.

The secondary structure was predicted by RNAfold as shown in (A). The triplet elements consist of the dot-bracket structure and the sequence content. (B) A moving window (length = 3) will slide over from the left end to the right end of the dot-bracket structure, and a sub-structure with the length of 3 will be extracted each time. Each sub-structure combined with the middle nucleotide of the three adjacent nucleotides will be counted as one triplet element. (C) Because there are only two structural statuses for each nucleotide (paired/bracket versus unpaired/dot), in total, three adjacent nucleotides will have 8 different structural combinations (e.g., “(((”, “((.”, “(..”, “(.(”, “.((”, “.(.”, “..(”, and “…”). With “A”, “G”, “C”, and “T” nucleotides, there are 32 triplet elements that have both sequence content (i.e., nucleotide) and structural information. So for each sequence, the number of 32 triplet elements will be counted and normalized through dividing by the total element number. The normalized data was shown in (D).