In 1986, the British police asked the molecular biologist Alec Jeffreys to apply a new genetic technique developed by him to the biological samples from a rape and murder case. In just a few years, DNA fingerprinting, which analyses small variable genomic regions, would become a common tool used throughout the world, both for law enforcement purposes and for paternity tests. At the beginning of this century, the sequencing of the human genome brought with it the possibility of analysing much larger segments, which has opened up a new era of applications for DNA profiles, not to mention boosting their popularity: according to Business Insider, on Black Friday 2017 one of the five most sold articles on the Amazon website was the genetic test kit from the Californian company 23andMe.
But what information can these tests provide? The most widespread use of DNA tests that are direct-to-consumer (DTC) is genealogical analysis, which determines from which region of the world the ancestors of an individual proceed based on certain genetic markers. According to MIT Technology Review, more than 12 million people have already undergone this type of test.
The year 2017 was the great boom for personal genetic tests, and in the heat of this growing market has emerged a multitude of companies that in certain cases are not limiting themselves to the innocuous results of an ancestry test, but are trying to extend their services towards the field of health. When NBC journalist Phil Rogers decided to try several DTC tests on himself, he discovered that different companies offered conflicting results about his genealogy, but he found a much bigger surprise: when he included samples from his dog Bailey, one of the companies informed him of his strong aptitude for sports such as boxing, basketball or cycling.
An active biomedical field
The relationship of certain genetic markers with factors associated with health is an active biomedical field. Programs like MyCode, launched in the state of Pennsylvania by the network of health centres Geisinger, scan the DNA of their patients in search of variants related to cancer or cardiovascular disease, so that those affected can undergo preventive treatments or adopt habits that help them minimize their risk.
This approach to personalized medicine is a great promise of current science, but regulatory bodies still greatly restrict these applications for DTC tests due to the risk of providing misleading information. However, given that genetic analysis businesses usually provide their clients with raw genomic data, some companies have already begun to take advantage of legal loopholes to offer a service for the interpretation of results.
In the eyes of many experts, these second opinions can do more harm than good: the company Ambry Genetics has published a study in which it has re-analysed a set of tests in which risky genes had been detected, discovering that 40% were false positives and that in other cases variants that had been labelled as high risk were actually not.
A possibility of predicting physical appearance
Predicting health risks by knowing the genomic sequence is a form of DNA phenotyping, or deducing observable traits from the genes. But an application of this new discipline that has recently captured the interest of scientists and the public is the possibility of predicting physical appearance. At the Catholic University of Louvain (Belgium), Peter Claes is working on 3D facial modelling from genomic data, and his group has recently identified at least 15 genome sites that influence facial morphology, especially in the nose. However, Claes believes that a faithful reconstruction of a face from DNA remains a long ways off. “There are still a lot of unknown factors both genetically as well as environmentally,” he tells OpenMind.
Claes’ prudent assessment stands in sharp contrast to the initiatives of other researchers. Last September, a study led by the biotechnology mogul J. Craig Venter presented a system that was supposedly able to identify the face of an individual from their genome. The results indicated that the algorithm used was correct 70% of the time when selecting the correct face from a collection of twenty.
However, the criticisms of Venter’s study have been strong, since once the faces with a different sex and ethnic group from those of the subject were discarded, the success rate dropped to 11%. “If they are first establishing the sex and ancestry of the DNA donor, then they have already gained the majority of the predictive value,” forensic geneticist Christopher Phillips of the Institute of Forensic Sciences at the University of Santiago de Compostela (Spain) explains to OpenMind. Phillips considers that “claims made for facial reconstruction techniques are both exaggerated and difficult to validate.”
Facial composites of suspects
Even bolder is the case of companies such as Parabon NanoLabs or Identitas, which offer DNA phenotyping services that include facial reconstructions. In fact, police forces in the US and abroad have already distributed facial composites of suspects generated by these companies, something that worries the experts. As Dennis McNevin, a forensic geneticist at the University of Technology in Sydney, Australia, tells OpenMind, “there is not much support in the literature for the proposition that an accurate facial composite can be produced from DNA at the current time.”
Thom Shaw, an IAI-certified forensic artist, performing a physical facial reconstruction and the digital adaptation of a Snapshot composite to reflect details gleaned from the victim’s facial morphology. Credit:
According to experts consulted by OpenMind, the problem with the services offered by these companies is that their methods haven’t been published, so the scientific forensic community hasn’t been able to validate them. “They operate what is essentially a black box,” says Phillips. For Claes, “Parabon is the best example of how not to bring this technology to market.” The researcher adds that the current claims about DNA phenotyping are also irresponsible, since an erroneous facial composite can divert a police investigation in the wrong direction.
But if we can overcome the current obstacles, will we someday be able to get a realistic image of a criminal from their DNA? “There will always be limitations to the prediction of faces from DNA,” says Claes. Not everything in one’s physical appearance is inheritable, advise the experts. McNevin adds that there are probably dependent traits from many different genes, and that understanding the interactions between them “could take decades.” Maybe one day parents will be able to know what the approximate appearance of their baby will be when they grow up, provided that, Claes concludes, the inflated proclamations of certain companies do not burn out the technology before it has had a chance to mature.