Performance comparison of Agilent new SureSelect All Exon v8 probes with v7 probes for exome sequencing

Abstract

Exome sequencing is becoming a routine in health care, because it increases the chance of pinpointing the genetic cause of an individual patient's condition and thus making an accurate diagnosis. It is important for facilities providing genetic services to keep track of changes in the technology of exome capture in order to maximize throughput while reducing cost per sample. In this study, we focused on comparing the newly released exome probe set Agilent SureSelect Human All Exon v8 and the previous probe set v7. In preparation for higher throughput of exome sequencing using the DNBSEQ-G400, we evaluated target design, coverage statistics, and variants across these two different exome capture products. Although the target size of the v8 design has not changed much compared to the v7 design (35.24 Mb vs 35.8 Mb), the v8 probe design allows you to call more of SNVs (+ 3.06%) and indels (+ 8.49%) with the same number of raw reads per sample on the common target regions (34.84 Mb). Our results suggest that the new Agilent v8 probe set for exome sequencing yields better data quality than the current Agilent v7 set.

Keywords: Agilent SureSelect; BGI; Enrichment quality; Exome sequencing; MGISEQ; NGS; Variant calling; WES.

Fig. 1

Venn diagram showing the intersection between the Agilent v7 exome (35.8 Mb), Agilent v8 exome (35.24 Mb), and Gencode V39 coding exons (34.93 Mb): A weighted, B unweighted

Fig. 2

Venn diagram showing the intersection of unique regions longer than 30 bp of the Agilent v7 exome (0.88 Mb) and Agilent v8 exome (0.39 Mb) with: A Gencode V39 coding exons (34.93 Mb); B NCBI RefSeq ALL exons + -20 bp (95.47 Mb)

Fig. 3

Barplots show the average values for on-targets, off-targets, duplicates, and unaligned reads in the samples from the v7 and v8 exome pools (downsampled results)

Fig. 4

A, B, C Performance of exome protocol in terms of coverage quality in downsampled samples: A Dependence of coverage quality of the v7 and v8 target regions from depth (mean ± SD values); B Graph showing the number of positions (bp, y-axis) vs. the coverage (x-axis); C Box plot showing the Fold-80 metric for the samples from the v7 and v8 pools

Fig. 5

Density plot showing Mean Depth vs. %GC content for: a Agilent v7; b Agilent v8. Density plot showing %GC content vs. mean depth. The data in this plot was collected by merging all samples from the V7 and V8 pools. Density estimation was performed using 2D plots. More specifically, we chose data points in a fixed rectangle (%GC content ∈ [0;1], mean depth ∈ [0;1000]) and split it into evenly spaced 200 × 100 grid and counted the data points in each cell of the grid. Finally, we normalised the grid to the range of [0,1] and plotted it using "jet" colormap from matplotlib library