Document Type

Poster

Abstract

The Ancestry Composition Algorithm is used by 23andMe to classify an individual's genetic information to inherited ancestry. Within this algorithm there are three specific steps to classify DNA: window classification, smoothing, and recalibration. Window classification uses kernel support vector machines to make initial ancestry classifications. Smoothing corrects for errors in the previously made classifications by using an autoregressive, pair hidden Markov model. Finally, the recalibration step accounts for biases by initially using an isotonic regression procedure; however, in cases of close geographic proximity, uses a hierarchical classifier to distinguish these ancestries.

html

Publication Date

2020

Keywords

Ancestry Composition Algorithm, kernel support vector machines, Markov model, genetic ancestry, 23andMe

Disciplines

Mathematics

Upload File

wf_yes

Share

COinS