Genome-wide genotyping remains a central tool in human genetics research, supporting applications such as genome-wide association studies (GWAS) and polygenic risk prediction. Traditional SNP array platforms, exemplified by the Illumina Infinium Global Screening Array-24 (GSA-24), have been widely adopted because of their relatively low cost, high reproducibility, and mature analytical workflows. When combined with genotype imputation, SNP arrays can capture a large portion of common human genetic variation.
More recently, targeted sequencing-based genotyping methods have emerged as alternatives to conventional arrays. The Twist Bioscience genome-wide SNP capture (GxS) platform applies hybridization-based enrichment to interrogate genome-wide SNP loci. Such sequencing-compatible capture assays may provide advantages in flexibility of assay design and integration into sequencing workflows.
This study was undertaken to provide an independent, third-party evaluation of the Twist GxS platform against whole-genome sequencing (WGS) on the same individuals. The authors report that prior to this work they were unable to find published third-party evaluations of Twist's genome-wide SNP capture platform.
The evaluation used two distinct cohorts organized as nuclear families. The Twist GxS platform was applied to 555 individuals comprising 184 nuclear families. Whole-genome sequencing genotype calls generated for these same 555 individuals were used as the benchmark standard for comparison.
Separately, the Illumina GSA-24 platform was evaluated on a different cohort of 987 individuals organized into 279 nuclear families. There is no sample overlap between the GxS cohort and the GSA cohort reported in this study.
The central comparison in this study is between genotype calls produced by the Twist GxS capture-based genotyping approach and genotypes obtained from WGS on the same individuals. WGS genotypes are treated as the reference or benchmark to which GxS performance is compared.
For broader context, the study also reports parallel evaluation of the Illumina GSA-24 platform on a separate set of families, enabling side-by-side contrasts of performance characteristics across commonly used genotyping technologies.
The authors indicate that key metrics for assessing the GxS platform include genotype concordance with WGS — that is, the agreement between genotype calls made from the capture assay and the WGS benchmark — and Mendelian violation frequency, a family-based quality metric that quantifies genotype inconsistencies with expected inheritance patterns.
These metrics were selected to capture both per-site/genotype-level agreement with the sequencing gold standard and family-structure integrity that affects downstream analyses in family-based genetic studies.
SNP arrays like the GSA-24 have historically been favored in many large-scale studies because they are cost-effective and supported by well-established analytical pipelines, including imputation to reference panels to increase variant coverage. Sequencing-based genotyping approaches, including hybrid-capture SNP targeting, aim to marry the advantages of sequencing (flexible workflows, potential to interrogate custom loci) with a lower per-sample cost than deep WGS.
The Twist GxS platform specifically uses hybridization enrichment to target genome-wide SNP loci, which can be advantageous for certain sample types and laboratory workflows. The authors cite recent work showing utility of the Twist GxS approach for challenging and degraded DNA samples, though they state that a comprehensive third-party benchmarking study was lacking before their work.
The preprint notes a gap in the published literature: apart from vendor reports and limited use-case demonstrations, the authors did not identify independent third-party evaluations of the Twist genome-wide SNP capture platform. This study aims to fill that gap by benchmarking GxS against WGS in a family-based design and by presenting comparative data relative to a commonly used SNP array platform.
The authors declared no competing interests. Funding for the reported work was provided by the Simons Foundation (SF497800).
Note on peer review and data availability
This report is presented as a preprint on bioRxiv and has not undergone peer review. The article includes links to additional information, metrics, and data/code resources as provided on the bioRxiv record. Specific numerical results for concordance rates, Mendelian violation counts, or other performance statistics are not reported in the portion of the source text provided here.