It's a common misconception that evolution has a sense of direction – a notion that biology nerds around the world are constantly trying to correct.

But new research reveals there may be a semblance of truth to this misconception, at least more than we ever realized.

While it's not as straightforward as mutation with a purpose, it now appears that not all DNA is equal when it comes to being mutable. At least not in a flowering roadside weed known as thale cress (Arabidopsis thaliana).

"We always thought of mutation as basically random across the genome," says plant scientist Grey Monroe from the University of California, Davis.

"It turns out that mutation is very non-random and it's non-random in a way that benefits the plant. It's a totally new way of thinking about mutation."

For a genetic mutation (or variant) to arise in the first place, several things have to go right for it. First, DNA must be altered within germline cells – cells that pass on their genetic material to an organisms' offspring.

This could involve a change to a single 'letter' in a DNA sequence, through UV damage, for example, or the loss of a gene, or the mix-up of an entire chromosome through errors when the genetic material is being copied and passed on.

Then this damage must elude several cellular mechanisms that are there to prevent such changes from being transferred. This includes DNA repair systems, or, for extreme mutations, programmed cell death (apoptosis).

If the mutation evades these processes, it can then be passed onto the next generation.

Most mutations that include the change of a single 'letter' are neutral, in that they don't give rise to any significant changes in the organism's form or function.

But for those that do cause changes, whether they continue through to the following generations can be subject to the whims of natural selection.

It's at this point that evolution was thought to do most of the sorting out between the good mutations and the duds. For example, if a mutation hinders the survival of a plant or animal, it is unlikely to remain for long.

While selection forces can constrain which mutations are passed on down through the generations, the mutation itself has generally been regarded as an unpredictable dice roll across the organism's genetic library.

"Since the first half of the twentieth century, evolutionary theory has been dominated by the idea that mutations occur randomly with respect to their consequences," the team writes in their paper.

Monroe and colleagues used the plant equivalent of a lab rat – the aforementioned thale cress – to test the assumption that mutation was indeed randomly distributed across a genome. They analyzed the genomes of 400 plant lines, and to their surprise this is not what their data showed.

Instead, they found certain regions of the plant's genome were far more prone to mutations than others.

"These are the really important regions of the genome," says Monroe. "The areas that are the most biologically important are the ones being protected from mutation."

This remained true whether they looked at coding or non-coding parts of the genetic code, suggesting the effect wasn't due to specific types of DNA, but the region as a whole.

"Evolution around genes in Arabidopsis appears to be explained by mutation bias to a greater extent than by selection," Munroe and team write, explaining that if this discrepancy was being caused later by natural selection, their analysis would have detected more unique gene variations than observed (as they would have been lost later in the process).

What's more, the data revealed epigenetic factors, like how the DNA is wound around certain proteins, and DNA repair mechanisms predict which parts of the genome are less prone to mutations. There was already strong evidence that DNA repair is targeted to active gene regions, which this study also supports.

Knowing how thale loads the dice when it comes to mutations could have implications not just for other plants, but for understanding evolution and disease in just about all species.

"It means we can predict which genes are more likely to mutate than others and it gives us a good idea of what's going on," says Weigel.

"This is exciting because we could even use these discoveries to think about how to protect human genes from mutation."

These findings suggest natural selection has skewed the probability of mutations across an organism's genetic library.

So while an individual mutation is indeed still random in terms of its consequence, the position across a genome is biased to favor an organism's survival, even before any possible effects of the mutation come into play in the game of natural selection.

Their research was published in Nature.