A genotype imputation reference panel specific for native Southeast Asian populations

Abstract

We report the development of a "Southeast Asian Specific (SEA-specific) Reference Panel" through a "Cross-panel Imputation" approach, consisting of 2550 samples derived from the GA100K, SG10K, and the Peninsular Malaysia Orang Asli (OA) datasets, covering 113,851,450 variants. The SEA-specific panel produced more high confidence variants than 1000 Genomes Project (1KGP) when imputing the OA (8.9 million SEA-specific vs 8.1 million 1KGP) and the Singapore Genome Variation Project (SGVP) (12.5 million SEA-specific vs 11.8 million 1KGP) genotyping datasets. Further, the SEA-specific panel imputed SNPs with better estimated quality scores (INFO, DR2 and R²) on the OA genotyping dataset when comparing with TOPMED and the Human Genome Diversity Project, but performed similarly on SGVP dataset. This panel also exhibited higher recall and non-reference disconcordance rates, indicating the influence of ancestry closeness of the reference panel. However, we note that the imputation accuracy may be compromised by the size of the reference panel.

Fig. 1. Average of internal quality score in different imputation scenarios.

The average internal quality score was generated in different allele frequency bins, in (a) SGVP dataset and (b) OA dataset. Imputation using the SEA-specific panel exhibited a better performance in both SGVP and OA genotyping datasets. c Internal quality score when imputation was performed on individual OMNI and AFFY genotyping datasets and subsequently merged. The dataset imputed prior to merging shows better performance than when imputation was performed on lower density of SNPs (imputed after merging of OMNI and AFFY genotyping data).

Fig. 2. Distribution of allele frequencies bins in imputed datasets.

The proportion of SNPs across allele frequency bins in the imputed genomes using different strategies. The SEA-specific reference panel consistently exhibited imputation of rare variants with lower frequencies (0-0.05) in all scenarios tested.

Fig. 3

Flow chart illustrating QC Criteria and merging procedure for genotyping datasets.

Fig. 4. Schematics of Imputation on the genotyping dataset and reference panel construction.

a The performance of pipelines containing a combination of reference panels and imputation tools was tested on the OA genotyping data. b Workflow of Reference Panel construction. Cross-panel imputation was employed to capture as many variants as possible from merged reference panels.