Fig 2: Measuring technical variation using DNA ladders.a Density distribution of k-mer counts for each cn unit (n = 570 k-mers per unit; 1 cn = red, 2 cn = yellow, 4 cn = green, and 8 cn = blue) in a simulated and experimental libraries (prepared with KAPA HyperPlus PCR-free and sequenced on Illumina HiSeq X Ten), with the mean, standard deviation and coefficient of variation indicated. For the experimental library, the n = 14 independent synthetic ladders where examined over 5 independent experiments with comparable results (see HiSeq X Ten/PCR-free libraries in Supplementary Figs. 5, 6, and 7a). b Modeling the impact of common NGS technical variables such as error rate, library depth, and library complexity on the regression slope of the synthetic DNA ladder. c The panel illustrates the variability associated with cn units across a range of libraries generated with different sequencing technologies (e.g., ONT Nanopore MinION, ONT Nanopore PromethION, Illumina HiSeq X Ten, Illumina NextSeq 500, and Illumina HiSeq 2500 instruments) or prepared with alternative protocols (e.g., KAPA, KAPA HyperPlus PCR-free/PCR-based and Nextera XT kits, or target enrichment by oligonucleotide hybridization). The y axis indicates k-mer counts, whilst the x axis indicates the position across the sub-sequence length (600 nt). The mean k-mer count (opaque line) and standard deviation (transparent line) are determined across all n = 14 independent synthetic ladders in all the sequencing experiments.

Fig 3: Improved coverage of human and bacterial genomes with Nextera DNA Flex. a Coverage across important regions of the human genome by three library preparation kits: Nextera DNA Flex, TruSeq Nano, and TruSeq PCR-free. b Coverage across extreme regions of the human genome by three library preparation kits: Nextera DNA Flex, TruSeq Nano, and TruSeq PCR-free. c Libraries generated from the small genomes of bacteria with low (B. cereus), medium (E. coli), and high (R. sphaeroides) GC content at 1 ng inputs. Less coverage variation was observed with libraries prepared by Nextera DNA Flex compared with libraries prepared by other commercially available library preparation kits (NEB, NEBNext Ultra DNA Library Prep Kit; Kapa A, Kapa HyperPlus Kits; Kapa B, Kapa HyperPrep Kits), particularly for the low and high GC content genomes of B. cereus and R. sphaeroides. The method of DNA fragmentation and whether PCR amplification was used during library preparation is indicated

Fig 4: Spurious variants are much more abundant in samples treated with fragmentase. (A) Design of the experiment testing different fragmentation and ER methods as well as DNA input source. Libraries were made from NA12814 and Hs27 DNA using either KAPA HyperPlus with ER mix 1 or 2 or KAPA HyperPrep with ER mix 1 (see the ‘Materials and Methods’ section). They were captured on the same probes as in Figure 2 and sequenced. (B) The variant incidence (excluding the expected variants) for the six resulting samples at single-strand and duplex consensus levels.