Skip to main content
. 2011 Sep 14;40(2):487–498. doi: 10.1093/nar/gkr629

Table 1.

Parameter list used for coaxial stacking and junction family type prediction

Feature Description
|L1|, |L2|, |L3| Loop lengths between junctions, labeled according to the 5′ to 3′ orientation of the entire RNA structure
Min(|L2|,|L3|), Min(|L1|,|L3|), Min(|L1|,|L2|) Minimum loop lengths
|L1′|, |L2′|, |L3′| Loop lengths sorted in ascending order (|L1′| ≤ |L2′| ≤ |L3′|)
A(L1), A(L2), A(L3) Maximum number of consecutive adenines in loops L1, L2 and L3, respectively
ΔG(H1,H2), ΔG(H2,H3), ΔG(H3,H1) Thermodynamic free-energy associated to the helical stacking between H1 and H2, H2 and H3, H3 and H1, respectively

List of the 15 parameters used for 3-way junctions. Similarly, for 4-way junctions, there are 18 parameters. For higher-order junctions, there are 10 parameters because we use a divide and conquer approach as described in the text. At the nodes of every decision tree, 3 parameters are selected randomly out of the total for 3-way and higher-order junctions in order to grow the tree by partitioning the node into 2 new branches, and 4 parameters for 4-way junctions.