国家神经系统疾病临床医学研究中心数据管理与共享系统

发表杂志：BMC Genomics

发表类型: Research Article

影响因子(IF): 4.4

PMID: 37353738 Pubmed原文链接 全文下载

The purpose of this study is to construct a pipeline and establish applicable fingerprint panels for sample tagging of whole genome bisulfite sequencing (WGBS) data using genotype information of whole genome sequencing (WGS). Hundreds of thousands of genomes from healthy people or patients have been sequenced by whole genome sequencing (WGS). Large-scale WGS projects that sequence at least tens of thousands of samples, such as Personal Genome Project UK (PGP-UK) and China Metabolic Analytics Project (ChinaMAP), should and have paid attention to the problem of sample mix-up, and the researchers solved it mainly by applying fingerprint genetic variants for sample tagging. The basic rationale is to tag each sample with a unique combination of fingerprint variants genotypes. Currently, quite a few panels of fingerprint variants have been proposed for sample tagging of DNA data by comparing genotypes of the fingerprint variants that are respectively generated by WGS and other methods or platforms. In contrast, very few panels have been proposed or optimized to check sample identities of multi-omics data, such as epigenomics data. Until now, only one panel consisted of 50 fingerprint SNPs has been published for sample tagging of transcriptomic data (Yousefi et al., 2018). PGP-UK applied the 65 control SNPs on Illumina HumanMethylation450 BeadChip array to tag whole genome bisulfite sequencing (WGBS) data and DNA methylation array data, but more details should be reported on genotype comparisons between mismatched pairs of WGS and WGBS data to convince the capability of the 65 SNPs. As WGBS is an increasingly accessible technique that is widely applied in medical and biological research, how to tag WGBS data is a question that should be systematically investigated. Taking the advantage of large-scale WGS and WGBS for stroke patients in the Third China National Stroke Registry (CNSR-Ⅲ), we constructed an optimized pipeline and designed a fingerprint panel of A/T polymorphic single nucleotide variants (SNVs) for sample tagging of WGBS data. To investigate their capability to tag WGBS data, exhaustive comparisons of WGS-based and WGBS-based genotypes were conducted within a first batch of 94 samples and then within a second batch of 240 samples. Moreover, to figure out the lower limit for the number of fingerprint variants in the panel that was capable to tag WGBS data using this pipeline, we also explored the performance of another 6 fingerprint panels. Taken together, this study systematically investigated sample tagging of WGBS data using genotypes of WGS, and provided a pipeline and a few applicable fingerprint panels. Their application would help to integrate multi-omics data of large-scale sequencing projects.

Bis-SNP was used in genotype calling from WGBS data. Compared with traditional DNA data tagging, optimization for WGBS data tagging included: 1) eliminating variants with wildtype homozygous and missing genotype; 2) reserving variants with identical genomic coordinates and reference/alternative alleles for genotype comparison. Exhaustive permutations of sample order were conducted that WGS-based and WGBS-based genotypes called from the identical sample and different samples were compared. By calculating genotype consistency rate, a self-designed autosome-wide A/T polymorphic SNV panel was tested to distinguish WGS-based and WGBS-based genotype calls of an identical sample from those of different samples in the first batch of 94 samples, and its capability was further tested in a second batch of 240 samples. Then, another 6 panels, including Affymetrix Genome-Wide Human SNP Array 6.0 and 5 fingerprint panels for tagging DNA data, were also tested among the 2 batches to identify the lower boundary for the number of fingerprint genetic variants for sample tagging of WGBS data.