My biology textbook mentions that DNA is antiparallel and it got me wondering - can DNA be parallel? What would happen if it was parallel? Could DNA still replicate correctly?
Interestingly the answer is yes. The problem with parallel strands is that the DNA is not pairing in the known way it does when it is combined anti-parallel (or Watson-Crick-pairs). I have found different images illustrating the problems:
Both images are from this blog post (originally from a publication cited there which is not available online). Both pictures show a distorted DNA structure.
This article ("NMR structure of a parallel-stranded DNA duplex at atomic resolution") shows a more schematic solution of this problem by showing the single pairings. C and G cannot build up 3 hydrogen bonds as they do this in the Watson-Crick-pairing.
Subfigure A and C show the Watson-Crick-pairs, the others come from parallel bindings.
This changes in structure disrupts the function of the DNA (which for example depends on the reverse strand for proof reading). Enzymes which replicate, transcribe, repair etc. DNA will not work on this changed structure, as their function depends on the structural features of the DNA.
I think one interesting possibility is that using PARALLEL construction it's very hard to make 'interesting' structures. This isn't important for DNA itself, but is critical for RNA molecules now... and even more so historically in the 'RNA world'. Consider: if making a structure using ANTIPARALLEL pairing, all the nucleic acid needs to do is 'turn around' 180 degrees (known as a 'hairpin' turn) before being able to pair with itself (i.e. form starting structure). This is well illustrated by tRNA http://www.wiley.com/legacy/college/boyer/0470003790/structure/tRNA/trna_intro.htm
Now--try to draw a tRNA-like structure using PARALLEL pairing--a strand must go through a FULL 360-degree turn before being able to self-pair. This is wasteful and perhaps even impossible without lots of added sequence. Since most folks accept that the beginnings of life were RNA or RNA-like molecules that were both functional (which requires meaningful shapes) as well as 'coding' (DNA or RNA sequence), if shapes were overly convoluted or could not form, life cannot start.
This would be an evolutionary, as opposed to structural, argument against parallel structures.
There is another issue here which is more biological. In general, the enzymes that replicate and "read" DNA run in one direction along the single strand of DNA that they are operating on. The fact that the other strand of DNA is running in the opposite direction makes it impossible for these enzymes to jump from one strand to the other and ensures that they are operating on the proper information to do their job. If the DNA strands were parallel, the enzymes would be likely to attach at a locus and then randomly read information from either strand, and the information produced would be gobbldygook. So while it is chemically possible to have parallel DNA, a biological system that was based on parallel DNA would be at a big competitive disadvantage and so if they ever arose, they were probably driven extinct by the systems we now see that use anti-parallel DNA.