Evaluation of consensus strategies for haplotype phasing

Abstract

Haplotype phasing is a critical step for many genetic applications but incorrect estimates of phase can negatively impact downstream analyses. One proposed strategy to improve phasing accuracy is to combine multiple independent phasing estimates to overcome the limitations of any individual estimate. However, such a strategy is yet to be thoroughly explored. This study provides a comprehensive evaluation of consensus strategies for haplotype phasing. We explore the performance of different consensus paradigms, and the effect of specific constituent tools, across several datasets with different characteristics and their impact on the downstream task of genotype imputation. Based on the outputs of existing phasing tools, we explore two different strategies to construct haplotype consensus estimators: voting across outputs from multiple phasing tools and multiple outputs of a single non-deterministic tool. We find that the consensus approach from multiple tools reduces SE by an average of 10% compared to any constituent tool when applied to European populations and has the highest accuracy regardless of population ethnicity, sample size, variant density or variant frequency. Furthermore, the consensus estimator improves the accuracy of the downstream task of genotype imputation carried out by the widely used Minimac3, pbwt and BEAGLE5 tools. Our results provide guidance on how to produce the most accurate phasing estimates and the trade-offs that a consensus approach may have. Our implementation of consensus haplotype phasing, consHap, is available freely at https://github.com/ziadbkh/consHap. Supplementary information: Supplementary data are available at Briefings in Bioinformatics online.

Fully exploiting SNP arrays: a systematic review on the tools to extract underlying genomic structure

“…In addition, some PRS tools need a previous imputation step, see the ‘Input Data’ column of the Tables for details. The phasing of the haplotypes, that is the separation between the maternally and paternally inherited copies of each chromosome, can be performed with tools like SHAPEIT 3, Eagle 2 and HAPI-UR [ 21 , 22 ]. Genotype imputation, the estimation of missing genotypes from a genotype reference panel, can be performed with tools like Minimac 4 from the Michigan Imputation Server, Impute 5, Beagle 5.2 or PBWT from Sanger Imputation Server [ 23 ].…”

eQTLHap: a tool for comprehensive eQTL analysis considering haplotypic and genotypic effects

“…Quality control was applied to the genotype datasets using PLINK to keep only SNPs with MAF > 0.01, Hardy-Weinberg equilibrium (HWE) < 10 −6 , the missing rate per individuals < 0.1, and missing rate per SNP < 0.1. Quality controlled genotype data were phased through consHap consensus phase estimator [23] by aggregating 15 applications of the SHAPEIT2 tool [24]. This approach has been reported to reduce the switch error rate significantly for datasets with small sample size (< 2,000).…”

A Pipeline for Phasing and Genotype Imputation on Mixed Human Data (Parents-Offspring Trios and Unrelated Subjects) by Reviewing Current Methods and Software

“…An imputation software employing reconstructed haplotypes requires the pre-imputation computational processing step of phasing to be performed [19]. Haplotypes can be reconstructed for both related and unrelated individuals.…”

“…The suggested overall sample size for unrelated individuals is over 50; the bigger the sample size is, the longer the haplotypes that can be reconstructed [21]. The recommended strategy for unrelated individuals is to perform phasing with haplotype frequency information from a reference population which has been more densely genotyped or sequenced [19,22,23], such as from the HapMap Project and the 1000 Genomes Project [24][25][26]. For related individuals, estimation is performed by considering both the haplotypes that are shared between family members and the haplotype frequency information of the reference population.…”