AG Algorithmen und Komplexität

This webservice is no longer available but eventually goes online at

Probability Profiling and Sampling of RNA Secondary Structures

... using a stochastic context-free grammar model

Based on our finding presented in paper "Markus Nebel, Anika Scheid. Evaluation of a Sophisticated SCFG Design for RNA Secondary Structure Prediction, Theory in Biosciences (2011) 130: 313-336," this web-service allows to compute probability profiles for given RNA sequences (primary structures). Such a profile consists of probabilites (given here in graphical representation) of each nucleotide to be unpaired in context of different loop-types (details see below). Thus it provides information on the expected folding of the given sequences. We have chosen tRNA data for this illustrative web-service for a user can most easily identify the common clover leaf structure e.g. by three peaks in the Hplot which 1-to-1 correspond with the three hairpin loops of the molecule. The service allows to oppose the probability model implied by our sophisticated stochastic context-free grammar (see paper) to the well-known partition function (PF) approach.

Primary structure to fold:
Folding Algorithm:


  • This webservice builds on probabilities derived from tRNA data. Others are available but have so far not been incooperated into this service.
  • In accordance with the partition function approach, isolated basepairs are allowed and hairpin-loops are restricted to a size of at least 3 nucleotides.
  • Primary structure needs to be entered as word over {A,C,G,U} (upper case only, no spaces).
  • Computation time is limited to two minutes. Inputs which need more time to be processed will be discarded.
  • As the result of a computation this webservice will provide plots of the probability for a specific nucleotide (identified by its position in the sequence) to be unpaired in context of different structural motives
    • hairpins (Hplot),
    • bulges (Bplot),
    • interior loops (Iplot),
    • multiloops (Mplot) or
    • exterior loops / tails (Eplot).
    Furthermore, the predicted folding (most likely most frequently sampled structure) will be provided by its graphical representation.
  • In order to be able to compare our results to those obtained by sampling based on partition functions, the corresponding algorithms are also provided by this webservice (choose the appropriate "Folding Algorithm").