Twist exome capture allows for lower average sequence coverage in clinical exome sequencing

Abstract

Background: Exome and genome sequencing are the predominant techniques in the diagnosis and research of genetic disorders. Sufficient, uniform and reproducible/consistent sequence coverage is a main determinant for the sensitivity to detect single-nucleotide (SNVs) and copy number variants (CNVs). Here we compared the ability to obtain comprehensive exome coverage for recent exome capture kits and genome sequencing techniques.

Results: We compared three different widely used enrichment kits (Agilent SureSelect Human All Exon V5, Agilent SureSelect Human All Exon V7 and Twist Bioscience) as well as short-read and long-read WGS. We show that the Twist exome capture significantly improves complete coverage and coverage uniformity across coding regions compared to other exome capture kits. Twist performance is comparable to that of both short- and long-read whole genome sequencing. Additionally, we show that even at a reduced average coverage of 70× there is only minimal loss in sensitivity for SNV and CNV detection.

Conclusion: We conclude that exome sequencing with Twist represents a significant improvement and could be performed at lower sequence coverage compared to other exome capture techniques.

Keywords: Exome sequencing; Genome sequencing; Uniformity of coverage.

Fig. 1

Comparison of exome kits and sequencing platforms. A Ratio of coding regions covered at ≥ 20× for different enrichment and sequencing platforms for RefSeq and Ensembl. B Boxplots of evenness scores for different enrichment kits and sequencing platforms. C Ratio of coding regions covered at ≥ 20× for different enrichment platforms when down-sampled to 50×. D GC content of insufficiently and sufficiently covered targets is significantly different for all kits and platforms (Mann–Whitney U-Test p value < 0.001)

Fig. 2

Comparison of enrichment kits and sequencing platforms at different coverage levels. A Overview of base pair coverage ratio at least 20× per platform for RefSeq coding regions. X-axis represents the mean coverage levels of the samples in each platform, y-axis represents the average ratio of base pairs that exceeds 20× coverage level for all samples in the corresponding kit/platform. B Boxplots represent the distribution number of coding variants for samples of each platform at different coverage levels. X-axis depicts the coverage levels, and y-axis shows the number of number of coding variants

Fig. 3

Comparison of Twist enrichment kit for 100× and 70× coverage levels. A Percentage of base pairs that exceeds 20× coverage level for OMIM genes (yellow) percentage of genes which were fully covered with at least 20× coverage (purple) B Venn diagram that represents the number of CNVs for samples enriched with TWIST at 100× (yellow) and 70× (purple) coverage levels