Granted Patent B1

US 12,354,710 · App. 18/157,595 · Granted Jul 8, 2025

Inventors: Chuong Do (Mountain View, CA); Eric Yves Jean-Marc Durand (Sceaux, FR); John Michael Macpherson (Santa Ana, CA); Brian Thomas Naughton (Mountain View, CA); Joanna Louise Mountain (Menlo Park, CA)

Assignee: 23andMe, Inc.

G16B40/00G06N5/04G06N7/01G06N20/00

Abstract

Claims (47)

1. A computer-implemented method comprising:

receiving genotype data of a user, wherein the genotype data comprises haplotype data for each autosomal chromosome and an X-chromosome of the user;

applying a pipelined process to the genotype data, wherein the pipelined process comprises:

dividing the genotype data into a plurality of windows, wherein each of the plurality of windows comprises a set of sequential single nucleotide polymorphisms (SNPs); and

applying a model to the plurality of windows to generate predicted ancestry composition information of the user for each of the plurality of windows;

receiving, via a user interface, a request to display ancestry composition information for the user;

displaying, via the user interface, ancestry composition information of the user in an interactive graphical user interface (GUI), wherein the displayed ancestry composition information is based at least in part on the predicted ancestry composition information, wherein the ancestry composition information is displayed using a chromosome-specific view, and wherein the chromosome-specific view comprises a list of the autosomal chromosomes of the user and the X-chromosome of the user that have undergone the pipelined process;

receiving, via the interactive GUI, a selection of a chromosome from among the autosomal chromosomes and the X-chromosome in the list; and

displaying, in the interactive GUI in response to the selection of the chromosome, geographical ancestry compositions of the selected chromosome, wherein displaying the geographical ancestry compositions of the selected chromosome comprises displaying a circle view of ancestries of the selected chromosome, and wherein the circle view comprises:

a first half-circle area illustrating a plurality of regions of ancestry inherited from a father of the user for the selected chromosome; and

a second half-circle area illustrating a plurality of regions of ancestry inherited from a mother of the user for the selected chromosome.

2. The method of claim 1 , wherein the haplotype data is phased haplotype data.

3. The method of claim 1 , wherein the haplotype data is unphased haplotype data, and wherein the pipelined process further comprises phasing the unphased haplotype data prior to applying the model.

4. The method of claim 1 , further comprising displaying, in the interactive GUI, a third partial-circle area, wherein the third partial-circle area:

surrounds at least a portion of the first half-circle area and illustrates one or more subregions of ancestry inherited from the father of the user for the selected chromosome; or

surrounds at least a portion of the second half-circle area and illustrates one or more subregions of ancestry inherited from the mother of the user for the selected chromosome.

5. The method of claim 1 , wherein the ancestry composition information comprises information pertaining to a proportion of the user's genotype data that is deemed to correspond to a geographical or ethnic ancestry.

6. The method of claim 1 , wherein the model is a machine learning model, a support vector machine (SVM), a neural network, or a Hidden Markov Model.

7. The method of claim 1 , wherein the circle view comprises different visual formats representing different ancestral continents.

8. The method of claim 7 , wherein the different visual formats comprise different colors or different patterns.

9. The method of claim 1 , wherein the model comprises a plurality of classifiers, and wherein each of the plurality of classifiers determines a likelihood that a window is one of two geographical or ethnic ancestries.

10. The method of claim 1 , wherein the ancestry composition information is also displayed using a list view, and wherein the list view shows proportions and corresponding continent names.

11. The method of claim 1 , wherein the pipelined process further comprises correcting errors in the predicted ancestry composition information of the user to generate corrected ancestry composition information, and wherein the displayed ancestry composition information of the user is further based on the corrected ancestry composition information.

12. The method of claim 11 , wherein correcting errors in the ancestry composition information of the user comprises:

providing the ancestry composition information of the user for each of the plurality of windows of the genotype data to an additional model; or

reducing noise; or

correcting phasing errors; or

correcting correlated prediction errors.

13. The method of claim 12 ,

wherein the additional model is a Pair Hidden Markov Model (PHMM) or an Autoregressive Pair Hidden Markov Model (APHMM); or

wherein:

the additional model comprises a plurality of error correction models; and

correcting errors in the ancestry composition information further comprises comparing output of the plurality of error correction models.

14. The method of claim 12 , further comprising generating the additional model by:

training one or more machine learning models using one or more training sets of data comprising individuals having admixed or unadmixed ancestries.

15. The method of claim 1 , wherein the first area is displayed side-by-side with the second area in the interactive GUI.

16. The method of claim 15 , wherein the first area is on a left side of the second area in the interactive GUI.

17. The method of claim 1 , wherein the ancestry composition information of the user comprises posterior probabilities that a proportion of the user's genotype data corresponds to a geographical or ethnic ancestry, and wherein the pipelined process further comprises recalibrating the posterior probabilities to generate the ancestry composition information.

18. The method of claim 1 , wherein the ancestry composition information of the user comprises posterior probabilities that a proportion of the user's genotype data corresponds to a geographical or ethnic ancestry, and wherein the pipelined process further comprises clustering posterior probabilities until the clustered posterior probability meets or exceeds a threshold value.

19. The method of claim 18 , wherein the clustering is performed by adding the posterior probabilities of geographical or ethnic ancestries associated with a geographical hierarchy.

20. The method of claim 1 , further comprising receiving, via the interactive GUI, a request for an expanded view corresponding to a selected geographical region of ancestry of the user.

21. The method of claim 1 , wherein the pipelined process further comprises generating the model by:

training one or more machine learning models using one or more training sets of data comprising individuals having four grandparents of a same specific ancestry.

22. The method of claim 21 , wherein each of the one or more machine learning models learns a relationship between DNA sequence information and one or two particular geographical or ethnic ancestry.

23. The method of claim 21 , wherein each of the one or more machine learning models learns a relationship between DNA sequence information and one particular geographical or ethnic ancestry compared to all other geographical or ethnic ancestries.

24. The method of claim 1 , wherein each window has a length that is an expected length of a recombination-free segment.

25. The method of claim 1 , wherein the model comprises a plurality of models, and wherein each model of the plurality of models generates a likelihood that each window of the plurality of windows corresponds to a different geographic or ethnic ancestry.

Assignments (5)

Continuity (5)

References Cited (400)