Figure 4
From: TurboFold: Iterative probabilistic estimation of secondary structures for multiple RNA sequences
Flowchart for the computation of extrinsic base pairing information for x
m
. The induced base pairing proclivity matrices, denoted by 
, are computed utilizing the base pairing probability matrices, 
(lower triangular matrices), and the posterior co-incidence probabilities {
c
Î (m,s)}. The extrinsic information, 
, is computed as the normalized weighted sum of the induced proclivity matrices. The lines in pairing probability matrices represent helices composed of pairs with relatively high pairing probability. Lines in co-incidence probability matrices represent relatively probable regions of sequence alignment.