Hello
I am trying to use RNAfold to predict the probability of each base in a mRNA is paired (with any base). I know I can extract the pairing probabilities for a base pair i,j from the dotplot PostScript file returned by RNAfold.
Now I wonder if I can get the probability P(i) that the ith based is paired by summing over P(i,j) for all bases j that have a nonzero probability to be paired with i. From the original McCasKill paper it seems that this is indeed the case.
However, when I examine the output of RNAfold I get situations like this one
112 120 0.045806981 ubox
112 124 0.020216035 ubox
112 141 0.003712513 ubox
112 146 0.005067031 ubox
112 174 0.023560836 ubox
112 177 0.992775000 ubox
Where summation will lead to P(i)>1 which of course makes no sense (this particular example is taken from the structure prediction for ENST00000342555, but I have many other similar examples).
So: are the pairing probabilities supposed to be mutally exclusive? If not, is there a way to calculate the probability P(i)?
Thanks in advance
