Scientists have long speculated that each individual’s microbiome may be unique and static enough so that it could be used for identification. This becomes very important for forensic investigations, which we have written about before, and also raises many ethical concerns regarding privacy during microbiome sampling and donation. Previously, most of the studies on this topic were not exhaustive enough to provide any firm conclusions. Last week though, Curtis Huttenhower’s group from the Harvard School of public health published a powerful, and statistically robust method for tracing the a microbiome back to its host. The study was published in the Proceedings of the National Academy of Sciences.
Using the Human Microbiome Project (HMP) database, the scientists used machine learning to construct a test for the most important conserved metagenomics traits after comparing individuals’ microbiomes over time. The algorithm depended on both 16s rRNA sequences, as well as whole genome sequencing (in addition to derivatives of the whole genome sequencing). The researchers note that the algorithm is not just looking for microbiome genes that are conserved over time, but rather microbiome genes that are conserved over time and unique amongst the population. Overall, they found that after a year, their algorithm could accurately identify 86% of people based on their stool samples, with very few false positives. Other sites on the body, like the skin, were less effective for identification, but it was feasible to use them.
This team definitively proved that a microbiome can be used to identify its host. They admit that full sequencing if the microbiome is necessary, but regardless, it is possible. This has all sorts of ethical and privacy concerns associated with it. For example, microbiome data that is made publically available, though anonymized, could be traced back to its donors. This could include information like STDs or other diseases. Another obvious application of this would be in forensics, and it probably wont be long before a case hinges on microbiome evidence.