Dear colleagues,

I am designing RNA molecular probes for FISH, and I want them to adopt a hairpin structure. I have checked for the possible secondary structure formation of some of the candidate sequences in different tools (i.e., OligoCalc, mFold, IDT OligoAnalyzer, KineFold), but the results greatly differ between them.

For instance, in OligoCalc, a given RNA sequence will be predicted to form a perfect hairpin structure (with a self-hybridizing stem and a loop), while in mFold it is predicted to have some basepairs self-hybridizing inside the loop, which is something undesirable in molecular beacons.

Which of these different tools is more robust in its predictions, and what is your experience with them? Where do these discrepancies arise from (is there any particular parameter I should consider)? Would you recommend me any other approach for assessing the hairpin formation of my candidate sequences?

Thanks a lot!