Biobank-scale inference of ancestral recombination graphs enables genealogy-based mixed model association of complex traits

Abstract

Accurate inference of gene genealogies from genetic data has the potential to facilitate a wide range of analyses. We introduce a method for accurately inferring biobank-scale genome-wide genealogies from sequencing or genotyping array data, as well as strategies to utilize genealogies within linear mixed models to perform association and other complex trait analyses. We use these new methods to build genome-wide genealogies using genotyping data for 337,464 UK Biobank individuals and to detect associations in 7 complex traits. Genealogy-based association detects more rare and ultra-rare signals (N = 133, frequency range 0.0004% - 0.1%) than genotype imputation from ∼65,000 sequenced haplotypes (N = 65). In a subset of 138,039 exome sequencing samples, these associations strongly tag (average r = 0.72) underlying sequencing variants, which are enriched for missense (2.3×) and loss-of-function (4.5×) variation. Inferred genealogies also capture additional association signals in higher frequency variants. These results demonstrate that large-scale inference of gene genealogies may be leveraged in the analysis of complex traits, complementing approaches that require the availability of large, population-specific sequencing panels.

Ultra-fast genome-wide inference of pairwise coalescence times

“…In addition, while the accuracy of Gamma-SMC is comparable to current methods, it could be further improved. Directions include incorporating genetic maps for inference; using more accurate demographic models when building the flow field; incorporating a post-hoc TMRCA normalization step [29]; and using the conditioned sample frequency spectrum [22, 27], which utilizes allele frequencies in inference. In terms of speed, GPUs may offer additional acceleration.…”

“…One exciting potential application of Gamma-SMC is in constructing an ARG. Indeed, the clear hierarchical structure evidenced in the pairwise posterior TMRCAs (Figure 3) reflects the genealogical tree structure at a site which, when inferred along the genome, gives rise to the tree sequence representation of an ARG [29]. Gamma-SMC may improve branch placement and timing in the weaving step of ARG building methods such as ARGweaver [24] and ARG-Needle [29], as the continuous nature of the posterior distributions of coalescence times in Gamma-SMC may help resolve ambiguities present in discrete time ap-proximations.…”

“…The length of these chains, measured by the (scaled) number of generations until coalescence, is called the coalescence time , or the time to the most recent common ancestor (TMRCA). Coalescence times reflect the genealogical and genetic history of a sample, and therefore have been used in many population and statistical genetic analyses, such as reconstructing demographic histories [15, 25, 27, 28], detecting selection signatures [11, 21, 22], genome-wide association studies [20, 29], genotype imputation [29] and more.…”

Extremely sparse models of linkage disequilibrium in ancestrally diverse association studies

“…The Li-Stephens model 15 is widely used for imputation 54 and phasing, 55 and it is also an important component of scalable genealogy inference methods like tsinfer and relate 18,19 . Large-scale genealogies, in turn, have enabled a variety of powerful methods in statistical and population genetics 26,27,56 . In this study, we leveraged genome-wide genealogies to derive LDGMs and to address computational challenges associated with LD and ancestral diversity.…”

“…Capitalizing on the limited number of common ancestral haplotypes at most loci, tree sequences provide a highly compact representation of human genetic data 18,20 . Tree sequences, and the closely related ancestral recombination graph, have enabled powerful new methods for understanding ancestral relationships [21][22][23] , measuring selection 19,24,25 , and analyzing complex traits 26,27 .…”