Uncategorized

23andMe can use Machine Learning to Determine Ancestry

23andMe has taken genetic testing mainstream. They analyze your genes to tell you about your ancestry and/or your health, depending on which package you select. You send them a sample of your saliva, and they send you results about your ancestry and possible health issues.

In this patent – US9213947B1 – 23andMe uses machine learning to help correspond ‘phased haplotype data’ to specific ancestries. As background, a ‘haplotype’, or haploid genotype, is a group of alleles in an organism that are inherited together from a single parent. ‘Phasing’ helps relate each haplotype to one of the parents (the mother or the father).

23andMe’s learning machine uses a ‘support vector machine’, which can be described as supervised learning models with associated learning algorithms that analyze data used for classification and regression analysis. In this case, the learning machine will rummage through your phased haplotype data (generated based on your saliva) and classify them. The classifications are used to relate each haplotype to specific ancestries. This mapping can help provide a picture of your ancestry.

So where you from?

Leave a Reply