Predicting the stability of profiling signals of small RNAs

Profiling is a process that finds similarities between different RNA secondary structures by extracting signals from the Boltzmann sampling. The reproducibility of profiling can be identified by the standard deviation of number of features among Boltzmann samples. We found a strong relationship between the frequency of each helix class and its standard deviation of the frequency upon repeated Boltzmann sampling. We developed a perturbation technique to predict the stability of these featured helix classes without the need for repeated Boltzmann sampling, with accuracy between 84% and 94%, depending on the type of RNA. Our technique only requires 0.2% of the computation time compared to one profiling process.