Why has the program option "Ignore unstable structures", known from version 2, vanished?
Unstable structures are secondary structures with a positive free energy, i.e. under normal conditions it is very unlikely that the molecule will take this fold. Although biologically well motivated, computationally this threshold seems to be rather arbitrary. Thus, we have not integrated it explicitly. Still, you can achieve the same effect -- reducing the energy sorted result list to structures with zero or negative free energies -- as in the older version by setting the relative deviation to 100% (or less) for computations with sub-optimal energy deviations.

Why has the program option "match shape", known from version 2, vanished?
The "match shape" program option in version 2 was an output filter. It did not effect the runtime of the program, just the output. We think you can easily achieve the same effect via your browsers "Find" (e.g. Ctrl+F) tool. Users with a *nix background may want to use the GREP command, to filter RNAshapes output.

Why has the program option "max loop length", known from version 2, vanished?
In older versions, option "max loop length" set the maximum lengths of the considered internal and bulge loops. The default value was 30 as in almost all other RNA folding programs. The restriction had a very slight influence on the calculated structure and shape probabilities, but could highly affect runtime. Our new underlying "Bellman's GAP" framework hardwires this value in the generated binaries for the sake of speed. For us it is very easy to change this variable and generate new binaries, but it would be too much overhead to offer different binaries on BiBiServ. We still use the default value 30. We would happily generate binaries with a different value on your personal request, should you really need them.

Why are there differences between ordering of structure probabilities and free energies for macrostate grammar?
The computation of free energies violates Bellman's principle of optimality, while computation of structure probabilities does not. Although, probabilities are just a liniear scaling of free energies, which cannot change ordering, we observe these little errors because of slightly differently defined search spaces, due to the violation of Bellman's principle. --> MFE does not satisfy bellman's principle