Effective preparation of Plasmodium vivax field isolates for high-throughput whole genome sequencing

Abstract

Whole genome sequencing (WGS) of Plasmodium vivax is problematic due to the reliance on clinical isolates which are generally low in parasitaemia and sample volume. Furthermore, clinical isolates contain a significant contaminating background of host DNA which confounds efforts to map short read sequence of the target P. vivax DNA. Here, we discuss a methodology to significantly improve the success of P. vivax WGS on natural (non-adapted) patient isolates. Using 37 patient isolates from Indonesia, Thailand, and travellers, we assessed the application of CF11-based white blood cell filtration alone and in combination with short term ex vivo schizont maturation. Although CF11 filtration reduced human DNA contamination in 8 Indonesian isolates tested, additional short-term culture increased the P. vivax DNA yield from a median of 0.15 to 6.2 ng µl(-1) packed red blood cells (pRBCs) (p = 0.001) and reduced the human DNA percentage from a median of 33.9% to 6.22% (p = 0.008). Furthermore, post-CF11 and culture samples from Thailand gave a median P. vivax DNA yield of 2.34 ng µl(-1) pRBCs, and 2.65% human DNA. In 22 P. vivax patient isolates prepared with the 2-step method, we demonstrate high depth (median 654X coverage) and breadth (≥89%) of coverage on the Illumina GAII and HiSeq platforms. In contrast to the A+T-rich P. falciparum genome, negligible bias was observed in coverage depth between coding and non-coding regions of the P. vivax genome. This uniform coverage will greatly facilitate the detection of SNPs and copy number variants across the genome, enabling unbiased exploration of the natural diversity in P. vivax populations.

Figure 1. Distribution by sample origin of A) parasite density, B) % ring stage parasites pre-culture, C) duration of ex vivo schizont maturation, and D) % schizont stage parasites post-culture.

Vertical lines indicate median of distribution.

Figure 2. Distribution of human DNA quantity (%) by sample origin and blood processing step.

Vertical lines indicate median of distribution.

Figure 3. Distribution of P. vivax yields (ng µl⁻¹ packed RBCs) by sample origin and blood processing step.

Vertical lines indicate median of distribution.

Figure 4. Distributions of Illumina Read Alignments against the P. vivax, Human, and P. falciparum Reference Genomes.

*Mixed-species confirmed by PCR. **Poor maturation in ex vivo culture. Note: % P. vivax reads does not include reads which cross-map against P. falciparum.

Figure 5. Sequence Depth Distributions across the P. vivax Sal-1 Reference Genome and in Coding and Non-coding Regions.

Samples = 22 independent, pure P. vivax isolates. CDS = coding sequence.

Figure 6. Sequence Breadth Distributions across the P. vivax Sal-1 Reference Genome and in Coding and Non-coding Regions.

Samples = 22 independent, pure P. vivax isolates. CDS = coding sequence.

Figure 7. Comparative Sequence Depth Distributions between the P. vivax and P. falciparum Genome in THA-009 and DRW-004.

Plots indicate the number of bases (y-axis) with a given sequence depth (x-axis) in different regions of the P. vivax (green) and P. falciparum (blue) genome. Whilst the P. vivax coding and non-coding sequences peak close to the nominal total sequence depth, the P. falciparum coding and non-coding sequences peak at moderately different coverage depths.