Skip to main content

Table 4 The robustness of target identification across samples of different GC-contents

From: On an enhancement of RNA probing data using information theory

 

GC-content

\(n=100\)

\(n=200\)

\(n=300\)

\({\mathbb {P}}(s\in \Omega ^*)\)

GC-rich

\(0.778 \pm 0.172\)

\(0.732 \pm 0.196\)

\(0.735 \pm 0.195\)

Uniform

\(0.768 \pm 0.178\)

\(0.742 \pm 0.192\)

\(0.751 \pm 0.187\)

AU-rich

\(0.773 \pm 0.176\)

\(0.735 \pm 0.195\)

\(0.749 \pm 0.188\)

\({\mathbb {P}}(s^*=s)\)

GC-rich

\(0.720 \pm 0.202\)

\(0.655 \pm 0.226 \)

\(0.674 \pm 0.220\)

Uniform

\(0.669 \pm 0.222\)

\(0.646\pm 0.229\)

\(0.706 \pm 0.208\)

AU-rich

\(0.677 \pm 0.219 \)

\( 0.655 \pm 0.226 \)

\( 0.701 \pm 0.210 \)

\({\mathbb {P}}(s^*=s\mid s\in \Omega ^*)\)

GC-rich

\(0.925 \pm 0.259\)

\(0.895\pm 0.309\)

\(0.917 \pm 0.299 \)

Uniform

\(0.871\pm 0.288\)

\(0.871 \pm 0.309\)

\(0.940\pm 0.277\)

AU-rich

\(0.876\pm 0.283\)

\(0.891 \pm 0.308\)

\(0.936 \pm 0.280\)

  1. We generate 1000 random sequences of length n with different GC-contents, where GC-rich sequences consist of \(30\%\) Gs, \(30\%\) Cs, \(20\%\) As and \(20\%\) Us; Uniform comprise \(25\%\) Gs, \(25\%\) Cs, \(25\%\) As and \(25\%\) Us; AU-rich contain \(20\%\) Gs, \(20\%\) Cs, \(30\%\) As and \(30\%\) Us. This process can be done by software such as GenRGenS [34]. For each sequence, we then generate (unrestricted) Boltzmann samples of \(N=2^{10}\) structures together with a target structure s. We compute the probabilities of identifying the target utilizing the ensemble tree. We display mean and standard deviation