I'm an undergraduate currently writing a thesis on RNA secondary structure prediction, specifically building upon analysis done by the RNAStructure and Unafold software packages by modifying certain algorithms here and there. I'm actually not a biologist, however. I've taken all of my classes in physics and computer science, and I sort of stumbled into my current research, and since it seemed to be a good fit for my skill set I stuck with it. However this leaves me curious of a motivation for my current work. Why is RNA secondary structure prediction important and what are some of its applications?
I can think of at least a dozen applications for which it would be useful to know the secondary structure of a given sequence of RNA off of the top of my head. In no particular order:
- Simulation/visualization of RNA
- RNA interference (RNAi)
- RNA-RNA interactions
- RNA-DNA interactions
- RNA-protein interactions
- Ribosomal protein expression
- Forced evolution of RNA aptamers
- Synthetic tRNAs with four and five base-pair codons
- Termination of transcription (the process that makes mRNA from DNA)
Basically anything to do with RNA, which is one of the fundamental building blocks of all living creatures.
Simulation/visualization of RNA
Secondary structure (2°) forms the second (yep) level in the standard 4 level hierarchy of macromolecular structure, and consists of short-range interactions between and among residues. In RNA, 2° is usually thought of as being loosely identical with base-pairing interactions. In order to do a proper simulation/visualization you'll probably also need at least the 3° information, but 2° is important too.
As an mRNA gets translated by a ribosome, in many cases it can form into a complex 3D structure that will affect the translation process. For example, some of the mRNAs that encode proteins related to the metabolism of metal ions have a riboswitch that will themselves bind to metal ions and inhibit/upregulate expression of the related protein. For various reasons it's usually extremely difficult to get the full 3D structure of a riboswitch, so getting to know the 2° remains an important source of information about them.
Knowing the secondary structure of nucleic acids is very useful in many cases when working with them. The simplest case, and probably the most often used case is when you order or synthesize a small RNA or DNA and actually don't want this to have any stable secondary structure. There are many methods where you use a small nucleic acid that can bind to a complementary part of another nucleic acid (hybridization).
For example in PCR, which is used to amplify small amounts of DNA, you use small DNAs called primers that bind to specific parts of a larger DNA template. If those primers can form stable secondary structures by themselves, those structures would have to be broken up before the primer can bind to the template. Such competing structures can drastically reduce the efficiency of such experiments and are something that people try to avoid, e.g. by predicting secondary structures before choosing primers.
There are also methods that use short nucleic acids with attached labels such as fluorophores. Those labeled probes bind to specific target sequences, and you can then e.g. determine with a microscope where those probes are bound. In all such techniques you don't really want secondary structures in your probes as they interfere with the binding to the actual target.
When RNA is introduced first, it is often said to be single-stranded and secondary structure is not even mentioned. But even when acting as messenger RNA, secondary structure of RNA does play a role. One mechanism for transcription termination works by forming a stem-loop within a certain range of stability, and predicting those is necessary to predict the site where transcription is terminated.
There are also many types of regulatory RNAs, this is a very active and hot topic at the moment. The most well known are probably RNA interference (siRNA and miRNA), but there are also many ribozymes and especially in bacteria also riboswitches. I won't try to list the different types of regulatory RNAs here, there is a large number of different classes that are known, and new ones are still discovered. To understand those it is often necessary to know the secondary or tertiary structure of the RNAs.
You can often at least create some hypotheses on how an RNA could work by looking at the secondary structure. In riboswitches for example you typically have an on- and an off-conformation, and those can often be explained by secondary structures.
While the secondary structure is often not enough to understand how a specific RNA works, you still start by predicting it in most cases. In many cases you would additionally use experimental data to verify the prediction (e.g. SHAPE or inline probing).
Determining the actual three-dimensional structure of RNA is pretty hard. How hard depends heavily on the exact RNA, but it can easily take years to do this. If you're lucky you can crystallize it easily, but RNAs with flexible parts tend to not crystallize well. In that case you can do NMR to determine the structure, but that has a rather severe size-limitation and can easily take years for larger RNAs.
In case of NMR, knowing the secondary structure is very useful in pretty much every case. You need to assign your signals to specific bases, and this is much easier if you know the secondary structure. You can also use NMR to support or disprove specific secondary structure predictions.
More reliable secondary structure prediction would be pretty nice, the current methods still fail often enough. Though there is probably a limit on how good you can get such methods for RNAs with extensive tertiary structure.