While predicting what someone’s face looks like based on a DNA sample remains extremely problematic, it is getting easier to use such a sample to filter the right face from a face database. That’s what an international team led by KU Leuven has shown in an article (“Facial recognition from DNA using face-to-DNA classifiers”) published recently in Nature Communications.
“Facial recognition from DNA refers to the identification or verification of unidentified biological material against facial images with known identity. One approach to establish the identity of unidentified biological material is to predict the face from DNA, and subsequently to match against facial images. However, DNA phenotyping of the human face remains challenging,” wrote the investigators, who included scientists from Penn State University and the University of Pittsburgh, among others.
“Here, another proof of concept to biometric authentication is established by using multiple face-to-DNA classifiers, each classifying given faces by a DNA-encoded aspect (sex, genomic background, individual genetic loci), or by a DNA-inferred aspect (BMI, age). Face-to-DNA classifiers on distinct DNA aspects are fused into one matching score for any given face against DNA.
“In a globally diverse, and subsequently in a homogeneous cohort, we demonstrate preliminary, but substantial true (83%, 80%) over false (17%, 20%) matching in verification mode. Consequences of future efforts include forensic applications, necessitating careful consideration of ethical and legal implications for privacy in genomic databases.”
Our physical appearance, including our face, is hardwired into our genetic material. Scientists have already identified multiple genes that determine the shape of our face—from the distance between our nostrils to the shape of our chin.
Still, that doesn’t mean that we can draw somebody’s face based on a DNA sample, explained electrotechnical engineer Peter Claes, PhD, of KU Leuven, the senior and corresponding author of the study. “We believe that the shape of our face is determined by thousands of genes, but also by the food we eat and other living conditions. Therefore, it is unlikely that we will ever be able to accurately predict a lifelike face from DNA alone.”
For forensic analysis and other applications, the ideal scenario would be to find a match between a DNA sample from a crime scene and someone in a DNA database. But those databases are limited. If the DNA doesn’t match anyone in the database, however, it could still be used to predict what the perpetrator’s face looks like and make a sketch. This sketch could then be compared to a database with the faces of known criminals, for example.
“This method mostly helps rule people out. In practice, we don’t usually get any further than a sort of reference face, such as ‘a European male,’” continued Claes. “That’s not much use to a forensic investigator. Unfortunately, while we are learning about more and more genes that determine certain aspects of our face, this does not yet sufficiently translate to a better match between the predicted face and the faces in the database.”
The researchers have now developed a reverse approach that works better: “Instead of going from DNA to face, we’re trying to go from face to DNA. Using special software, we measure each face and check if this face is a possible outcome based on a unique bit of DNA,” according to Claes. “It then becomes a game of ‘guess who’. If the face is male and the DNA says it is a woman, all the men are eliminated. If the hair is blond and the genetic material confirms this, that eliminates all other hair colors.
“The more genes we identify, the more accurate this method becomes, and it will only continue to improve as our knowledge of the relevant genes grows.”
So what if the face you’re looking for is not in the database? “Then you will at least have a list of faces that look a lot like the one you are looking for. Instead of a single reference face, such as ‘Asian female’, you will have a list of Asian females who are most similar to the person you want to find,” explained Claes.
This requires large face databases, such as ID card or driver’s license databases. So, of course, there is an ethical-legal matter to consider, said Claes. “Working with databases that contain private information such as DNA or faces requires strict supervision to prevent misuse.”