Many of the promising applications of genome sequencing lie in associating genotypes (the unique genome set of a person) with physical and disease traits. However, there’s a newly-developed technology that uses genetic data to predict phenotypes (observable qualities that are affected both by the genotype and the environment) like facial features, skin and eye color. After the study’s publication in the Proceedings of the National Academy of Sciences, controversies on privacy issues got raised by the company behind it, Human Longevity, Inc.
Models that use genetic information for phenotype predictions are already existent. HLI researchers, led by genome-sequencing pioneer Craig Venter, recently made their own by building an algorithm to predict 3D facial structures and consequently produce face images. The HLI team claims accurate predictions for simple traits (like eye color, skin color and sex) but not so much for the more complex ones.
To do the procedure, HLI sequenced genomes of 1,061 San Diego residents with diverse ethnic backgrounds. Ages of the participants also varied, from 18 to 82 years old, with age averaging at 36. Combining genetic data and photographs, the team sought DNA factors that were linked with facial features (like cheekbone height), age, vocal qualities, height, weight, body mass index and eye color.
To test the validity of their algorithm, the researchers attempted to re-identify some individuals from the same pool. The process was able to correctly identify 8 out of 10 individuals in the ethnically mixed group. However, lower accuracy was achieved in predicting single-race participants—it was 5 out of 10 for African Americans and Europeans.
While there are limitations in terms of a group size when small, researchers believe that the predictions are sound. Obviously, one big application of this new technique could be in the forensic sciences. Through blood or any biological sample, it is going to be easier to perform identifications and link an extracted DNA for intelligence and law enforcement purposes.
HLI, however, is more concerned over the aspect of genome privacy protections. As a highly accurate facial prediction technology which is quite possibly going to be widely used in the near future, the company says it is important to set stricter guidelines for protecting genome data. Many online databases are public and HLI raised the risk of identifying anyone through DNA alone. In a statement, the company wrote, “A core belief from the HLI researchers is that there is now no such thing as true deidentification and full privacy in publicly accessible databases.”
But this stand from HLI raised controversies, which other experts in the field had criticized. One view is that building a restricted access is seen as a conflict of interest; HLI itself is a for-profit company that aims to build the largest database of human genetic information. “I think genetic privacy is very important, but the approach being taken is the wrong one,” said Jason Piper, co-author and a former employee of HLI. “In order to get more information out of the genome, people have to share.”
Other experts say that there is no need to be alarmed as HLI’s study is inaccurate. Computational biologist Yaniv Erlich published a paper criticizing HLI’s study, noting that the procedure did not use the power of whole-genome markers, used wrong metrics and did not truly identify anyone. HLI released a rebuttal saying that it “stands by the protection of genome data and the promotion of modern solutions for data exchange”. The company added that the paper was intended to encourage the discussion on sharing genetic information without compromising a person’s privacy.