Forensic DNA Phenotyping is of significant importance to the #lawenforcementforces, especially in solving #coldcases. This technique uses genes that affect the appearance to predict the ethnicity, race, and physical features of an individual i.e. eye color, skin color, hair color, and age. Reductively speaking, #DNAphenotyping reconstructs an individual’s face using his/her DNA. This face is generally matched to a #mugshotdatabase. Sometimes, the DNA is also matched to #publicgenealogydatabases, which are used by people to build #familytrees. From this, arises an ethical quagmire.
Nowadays, more and more people are sending their DNA samples to private companies like #23andMe and #Ancestry for identifying their ancestral origins, for getting information regarding their genetic #predispositiontocertaindisease(s), or in order to discover lost relatives. These companies sometimes share “anonymised” data by removing the names of the owners. But, can privacy be ensured if the owner’s face can be predicted.
Furthermore, the #traitpredictions currently in use are based on predictive models that have predominantly been developed using #modernEuropeanpopulations, so it would be prudent apply caution when analyzing non-European populations. At the current stage, there is a significant degree of error associated with this method; this is despite the fact that DNA phenotyping is advancing rapidly with the application of #machinelearningapproaches. However, this might be attributed to procedures regarding the collection and use of DNA in #criminalinvestigations, where it is susceptible to #humanerror, #humanbias, and abuse.
This brings us to another big issue: #racialbias. The fact that ancestry, ethnicity, and race are inherently fraught topics makes one wonder about the impact of the DNA phenotyping results on the general public. Making these results public would probably result in linking of certain communities with crime, i.e. ‘it’s in their DNA’, mindset.
Does this mean that mugshots based on DNA are useless? Not quite. The science underlying this technology is still in its infancy; meaning, using it in criminal cases would be hasty- law enforcement agencies should consider waiting a little longer before splurging on this expensive tech. At best, the phenotyping results will be helpful. At worst, they'll simply be another form of #racialprofiling.Tags: ancestry, criminal investigations, human bias, human error, machine learning approaches, trait predictions