Creating Forensic Sketches from DNA

This seems really science fictional:

It's already possible to make some inferences about the appearance of crime suspects from their DNA alone, including their racial ancestry and some shades of hair colour. And in 2012, a team led by Manfred Kayser of Erasmus University Medical Center in Rotterdam, the Netherlands, identified five genetic variants with detectable effects on facial shape. It was a start, but still a long way from reliable genetic photofits.
To take the idea a step further, a team led by population geneticist Mark Shriver of Pennsylvania State University and imaging specialist Peter Claes of the Catholic University of Leuven (KUL) in Belgium used a stereoscopic camera to capture 3D images of almost 600 volunteers from populations with mixed European and West African ancestry. Because people from Europe and Africa tend to have differently shaped faces, studying people with mixed ancestry increased the chances of finding genetic variants affecting facial structure.
Kayser's study had looked for genes that affected the relative positions of nine facial "landmarks", including the middle of each eyeball and the tip of the nose. By contrast, Claes and Shriver superimposed a mesh of more than 7000 points onto the scanned 3D images and recorded the precise location of each point. They also developed a statistical model to consider how genes, sex and racial ancestry affect the position of these points and therefore the overall shape of the face.
Next the researchers tested each of the volunteers for 76 genetic variants in genes that were already known to cause facial abnormalities when mutated. They reasoned that normal variation in genes that can cause such problems might have a subtle effect on the shape of the face. After using their model to control for the effects of sex and ancestry, they found 24 variants in 20 different genes that seemed to be useful predictors of facial shape (PLoS Genetics, DOI: 10.1371/journal.pgen.1004224).
Reconstructions based on these variants alone aren't yet ready for routine use by crime labs, the researchers admit. Still, Shriver is already working with police to see if the method can help find the perpetrator in two cases of serial rape in Pennsylvania, for which police are desperate for new clues.

If I had to guess, I'd imagine this kind of thing is a couple of decades away. But with a large enough database of genetic data, it's certainly possible.