Evaluating the Accuracy of Imputation Methods in a Five-Way Admixed Population

Motivation: Over the last few years, methods based on suffix arrays using the Burrows–Wheeler Transform have been widely used for DNA sequence read matching and assembly. These provide very fast search algorithms, linear in the search pattern size, on a highly compressible representation of the dataset being searched. Meanwhile, algorithmic development for genotype data has concentrated on statistical methods for phasing and imputation, based on probabilistic matching to hidden Markov model representations of the reference data, which while powerful are much less computationally efficient. Here a theory of haplotype matching using suffix array ideas is developed, which should scale too much larger datasets than those currently handled by genotype algorithms. Results: Given M sequences with N bi-allelic variable sites, an O(NM) algorithm to derive a representation of the data based on positional prefix arrays is given, which is termed the positional Burrows–Wheeler transform (PBWT). On large datasets this compresses with run-length encoding by more than a factor of a hundred smaller than using gzip on the raw data. Using this representation a method is given to find all maximal haplotype matches within the set in O(NM) time rather than O(NM 2) as expected from naive pairwise comparison, and also a fast algorithm, empirically independent of M given sufficient memory for indexes, to find maximal matches between a new sequence and the set. The discussion includes some proposals about how these approaches could be used for imputation and phasing. Availability: http://github.com/richarddurbin/pbwt Contact: richard.durbin@sanger.ac.uk Show more

A current approach to mapping complex-disease-susceptibility loci in genome-wide association (GWA) studies involves leveraging the information in a reference database of dense genotype data. By modeling the patterns of linkage disequilibrium in a reference panel, genotypes not directly measured in the study samples can be imputed and tested for disease association. This imputation strategy has been successful for GWA studies in populations well represented by existing reference panels. We used genotypes at 513,008 autosomal single-nucleotide polymorphism (SNP) loci in 443 unrelated individuals from 29 worldwide populations to evaluate the "portability" of the HapMap reference panels for imputation in studies of diverse populations. When a single HapMap panel was leveraged for imputation of randomly masked genotypes, European populations had the highest imputation accuracy, followed by populations from East Asia, Central and South Asia, the Americas, Oceania, the Middle East, and Africa. For each population, we identified "optimal" mixtures of reference panels that maximized imputation accuracy, and we found that in most populations, mixtures including individuals from at least two HapMap panels produced the highest imputation accuracy. From a separate survey of additional SNPs typed in the same samples, we evaluated imputation accuracy in the scenario in which all genotypes at a given SNP position were unobserved and were imputed on the basis of data from a commercial "SNP chip," again finding that most populations benefited from the use of combinations of two or more HapMap reference panels. Our results can serve as a guide for selecting appropriate reference panels for imputation-based GWA analysis in diverse populations. Show more

The worldwide burden of tuberculosis (TB) remains an enormous problem, and is particularly severe in the admixed South African Coloured (SAC) population residing in the Western Cape. Despite evidence from twin studies suggesting a strong genetic component to TB resistance, only a few loci have been identified to date. In this work, we conduct a genome-wide association study (GWAS), meta-analysis and trans-ethnic fine mapping to attempt the replication of previously identified TB susceptibility loci. Our GWAS results confirm the WT1 chr11 susceptibility locus (rs2057178: odds ratio = 0.62, P = 2.71e(-06)) previously identified by Thye et al., but fail to replicate previously identified polymorphisms in the TLR8 gene and locus 18q11.2. Our study demonstrates that the genetic contribution to TB risk varies between continental populations, and illustrates the value of including admixed populations in studies of TB risk and other complex phenotypes. Our evaluation of local ancestry based on the real and simulated data demonstrates that case-only admixture mapping is currently impractical in multi-way admixed populations, such as the SAC, due to spurious deviations in average local ancestry generated by current local ancestry inference methods. This study provides insights into identifying disease genes and ancestry-specific disease risk in multi-way admixed populations. Show more