Discovering population structure from patterns of identity-by-descent
Described are techniques for determining population structure from identity-by-descent (IBD) of individuals. The techniques may be used to predict that an individual belongs to zero, one or more of a number of communities identified within an IBD network. Additional data may be used to annotate the communities with birth location, surname, and ethnicity information. In turn, these data may be used to provide to an individual a prediction of membership to zero, one or more communities, accompanied by a summary of the information annotated to those communities.
1. A computer-implemented method comprising:
accessing a genotype of an individual;
generating a plurality of identity-by-descent (IBD) estimates between the individual and each of a number of reference panel individuals who are members of one of a plurality of reference panels, each of the reference panel individuals associated with one of a plurality of different communities of an IBD network;
the communities comprising a plurality of nodes, the nodes representing individuals and their associated genotypes, the nodes connected via edges with weights based on IBD estimates between the associated genotypes, the nodes grouped into the communities based on the edge weights;
inputting the IBD estimates into a set of models each generating a score for classifying the individual to one or more of the communities, each model corresponding to one of the communities and comprising a set of features defined from the IBD estimates shared among the individuals in the reference panels who are associated with the one of the communities, each set of features included in a model different from sets of features included in other models in the set of models;
sending a report summarizing the communities for which the individual's score exceeds a threshold.
2. The computer-implemented method of claim 1 , wherein an IBD estimate corresponds to a total amount of one or more chromosomes that is inherited from an ancestor that is common to the individual and one of the individuals in one of the reference panels.
3. The computer-implemented method of claim 1 , wherein the features are each defined as a number of connections between the individual and the individuals in one of the reference panels within specified ranges of total amounts of estimated IBD.
4. The computer-implemented method of claim 3 , wherein the features include separate features for each of the specified ranges for each of the different reference panels.
5. The computer-implemented method of claim 1 further comprising:
generating the IBD network by:
generating the plurality of IBD estimates, each IBD estimate being between each of a pair of individuals in one of the reference panels;
mapping the IBD estimates into edge weights:
applying a network clustering algorithm to a network defined by these edge weights to identify two or more communities; and
classifying the individuals to each community based on the IBD estimates.
6. The computer-implemented method of claim 5 wherein mapping the IBD estimates into edge weights further comprises:
choosing a target range of generations;
simulating a distribution of estimated IBD corresponding to relationships in which common ancestors fall within the target range of generations; and
defining the mapping based on IBD estimates associated with pairs of individuals with common ancestors who fall within the target range of generations.
7. The computer-implemented method of claim 5 wherein the network clustering algorithm is implemented as a recursive application of a modularity-maximizing community detection method.
8. The computer-implemented method of claim 5 further comprising:
refining the reference panels by, for each of a plurality of the individuals in the reference panel:
applying the network clustering algorithm a plurality of times;
measuring a stability ranking of one of the individuals in one of the reference panels by counting a number of times that the one of the individuals is assigned to a same community; and
maintaining the one of the individuals in the one of the reference panels if the stability ranking is greater than a threshold stability ranking.
9. The computer-implemented method of claim 1 , further comprising:
accessing pedigree information regarding the reference panel individuals, the pedigree information specifying information about ancestors of the reference panel individuals.
10. The computer-implemented method of claim 9 , further comprising:
annotating the communities with at least one of ancestral birth location labels and surname labels based on the pedigree information of the reference panel individuals associated with each community; and
wherein sending the report further comprises a summary based on at least one of the birth location labels and the surname labels for the communities when the individual's score exceeds the threshold.
11. The computer-implemented method of claim 9 , further comprising:
wherein accessing the pedigree information regarding one of the reference panel individuals comprises:
receiving the pedigree information, the pedigree information comprising at least one from the group consisting of:
a name of an ancestor of the reference panel individual,
a gender of the ancestor,
a birth location of the ancestor,
a date of birth of the ancestor,
a date of death of the ancestor, and
a surname of the ancestor;
filtering the pedigree information based on one or more criteria, the criteria including at least one from the group consisting of:
a recorded a birth date of the ancestor
a recorded death date of the ancestor, and
a recorded gender of the ancestor.
12. The computer-implemented method of claim 1 , further comprising:
generating ethnicity estimates for the reference panel individuals;
annotating the communities with ethnicity labels based on the ethnicity estimates associated with the reference panel associated with each community; and
wherein sending the report further comprises sending a summary based on the ethnicity labels for the communities when the individual's score exceeds the threshold.