Universal human papillomavirus typing by whole genome sequencing following target enrichment: evaluation of assay reproducibility and limit of detection

Abstract

Background: We recently described a method for unbiased detection of all known human papillomaviruses (HPV) types with the potential for the determination of their variant and integration from the resulting whole genome sequence data. Considering the complex workflow for target-enriched next generation sequencing (NGS), we focused on the reproducibility and limit of detection (LOD) of this new universal HPV typing assay in this study.

Results: We evaluated the reproducibility and LOD for HPV genotyping based on our recently published method that used RNA-baits targeting whole genomes of 191 HPV types, Agilent SureSelect protocol for target enrichment and Illumina HiSeq 2500 for sequencing (eWGS, enriched whole genome sequencing). Two libraries, prepared from pooled plasmids representing 9 vaccine HPV types at varying input (1-625 copies/reaction), were sequenced twice giving four replicates for evaluating reproducibility and LOD. eWGS showed high correlation in the number of reads mapped to HPV reference genomes between the two flow-cell lanes within (R² = 1) and between experiments (R² = 0.99). The number of mapped reads was positively correlated to copy number (β = 13.9, p < 0.0001). The limit of blank (LOB) could be calculated based on mapped reads to HPV types not included in each sample. HPV genotyping was reproducible for all 9 types at 625 copies using multiple cut-off criteria but LOD was 25 copies based on number of reads above LOB even when multiple types were present. eWGS showed no bias for HPV genotyping under single or multiple infection (p = 0.16-0.99).

Conclusions: The universal eWGS method for HPV genotyping has sensitivity, competitive with widely used consensus PCR methods with reduced type competition, and with the potential for determination of variant and integration status. The protocol used in this study, using defined samples varying in complexity and copy number, analyzed in replicate and duplicate assays, is applicable to most WGS methods.

Keywords: HPV typing; LOD; NGS; Reproducibility; Target enrichment.

Fig. 1

Experimental design for evaluation of reproducibility and limit of detection (LOD) of eWGS. To evaluate reproducibility of results, individual indexed libraries were prepared from 16 samples of defined HPV composition on two occasions 10 days apart (Experiments 1 and 2 in Fig. 1) resulting in 2 pooled libraries. Each library was enriched through hybridization with HPV RNA bait and each enriched library was sequenced on two flow cells. Thus 4 replicate results were obtained for each sample, encompassing experimental replicates (reproducibility of producing enriched library) and sequencing replicates (1a, 1b and 2a, 2b). As each defined sample was a pool of 4 to 5 HPV types with copy number ranging from 625 to 1 (composition shown in Table 1), the limit of detection could be assessed from the replicate results

Fig. 2

Evaluation of reproducibility in terms of number of reads mapped to expected HPV genomes. a reproducibility between 2 flow cell lanes for experiment 1 and experiment 2, and (b) reproducibility between experiment 1 and 2 in term of number of mapped reads (mean value of 2 replicates)

Fig. 3

Mapped reads to expected HPV genomes showing the relationship between copy number and (a) mean number of mapped reads, b fraction of reference genome, and (c) mean CV for the 9 HPV plasmids in the number of mapped reads in relation to copy number, and (d) mean CV for the HPV plasmids for the fraction of reference genome covered in relation to HPV copy number. Error bars indicates standard deviation