Lecture 2: Metadata

Lecture 2 Metadata: combining genomic data with other forms of data is uniquely powerful ● 15-year-old tracked his anonymous sperm donor biological father by submitting a saliva sample to FamilyTree.com that narrowed it down to 2 men. Then used Omnitrace.com to enter personal info about him to then find him ○ Problematic: Anonymous sperm donation isn't actually anonymous. No such thing as closed adoption anymore De-Anonymization: people who are anonymous in genetic studies can be easily identified ● Ex. Yaniv Ehrlick published no names in his research but used cross referenced Y chromosome data that can still be identified ○ Some researchers say that genetic data should be central data-hosting agencies so that it's off-limits from the public ○ Others say that participants should understand how their data could be used in the future before agreeing Despite rules and regulations about privacy, there may never be true genome anonymity Combining Genomic Data with Genealogies The Golden State Killer: lack of genetic privacy when combined with other metadata (age, location, genealogy) ● Ex. 15-year-old used genetic database, GEDMatch, which collects DNA from users across services to form a connection of family networks without de-anonymization algorithms ● 1970s-1980s sadistic rapes/murders unsolved but was solved when identified Mr.DeAngelo who had matched DNA with 2 unsolved murders in Ventura County, CA ○ Used crime scene DNA to construct a genetic profile and compare to other sources (GEDMatch). Identify relatives, identify, and the suspect. Retain the suspect and obtain discarded DNA samples to test if its a match at the crime scene ● DNA can narrow identity to 1 in 20 people from a database of 1.3 million, even if the person's DNA isn't in the databse ○ identify 60% of white Americans ○ GEDMatch had 1 million records (0.5% of population) If 2% of population was sampled, then the probability to be found is 90% ■ Due to genealogy -> allows you to identify unsampled individuals, combined with other DNA Databases and genelogical networks and our ability to identify/learn about genomes not in the databases will grow -> most individuals are already identifiable using these methods ● (2021) 29 bills across 16 states relating to genetics and insurance underwriting 23andMe Privacy Policy: states that it won't share your info to any public databases, insurance companies, or employers but yes to service providers (won't violate federal law but is a data broker) but will prptect your identify in the event of a data breach ● Data breach: claimed to be attackers randomly guessing the login info so data isn't shared but login data is ○ Claimed to be a targeted attack on Ashkenazim (Chinese customers) ○ Showed that their data isn't perfectly safe What can you do? unique passwords, request to delete your data, don't share genetic info Take Home Messages: ● 23andMe.com and ancestry.com were the industry leaders, but there were many DTC services out there. ● Services used SNP data to trace ancestry and to correlate genetic background with phenotypic traits. ● The FDA attempted to regulate genetic testing, as it would any medical test. ● 23andMe moved into formal genetic testing, research and drug development, and finally as a data broker. ● Although personalized genomics and personalized medicine will become a significant factor in health care, the SNP based firms are not keeping up. ● Ancestry.com leveraged its large genealogical database and combining it with genetics. Large genetic databases combined with other data present opportunities for law enforcement, but a serious challenge to privacy. ● Using relatives to get information will be an exponentially powerful tool, one that requires legislative action. ● Other genetic privacy issues abound. Genetic data are not anonymous. ● concerns overlap broadly with privacy in other disciplines,not easy to opt out.