Optimization of parasite DNA enrichment approaches to generate whole genome sequencing data for Plasmodium falciparum from low-parasitemia samples

Background: Owing to the large amount of host DNA in clinical samples, generation of high-quality Plasmodium falciparum whole genome sequencing (WGS) data requires enrichment for parasite DNA. Enrichment is often achieved by leukocyte depletion of infected blood prior to storage. However, leukocyte depletion is difficult in low-resource settings and limits analysis to prospectively-collected samples. As a result, approaches such as selective whole genome amplification (sWGA) are being used to enrich for parasite DNA. However, sWGA has had limited success in generating reliable sequencing data from low parasitemia samples. In this study, we evaluated whether enzymatic digestion with MspJI prior to sWGA improved genome coverage compared to sWGA alone. We also examined the potential of sWGA to cause amplification bias in polyclonal infections. Methods: DNA extracted from lab-created dried blood spots was treated with a modification-dependent restriction endonuclease, MspJI, and filtered via vacuum filtration. Samples were then selectively amplified using a previously reported sWGA protocol and subjected to WGS. Genome coverage statistics were compared between the optimized sWGA approach and the previously reported sWGA approach performed in parallel. Differential amplification by sWGA was assessed by comparing WGS data generated from lab-created mixtures of parasite isolates, from the same geographical region, generated with or without sWGA. Results: MspJI digestion did not enrich for parasite DNA. Samples that underwent vacuum filtration (without MspJI digestion) prior to sWGA had the highest parasite DNA concentration and displayed greater genome coverage compared to MspJI+sWGA and sWGA alone, particularly for low parasitemia samples. The optimized sWGA (filtration + sWGA) approach was successfully used to generate WGS data from 218 non-leukocyte depleted field samples from Malawi. Sequences from lab-created mixtures of parasites did not show evidence of differential amplification of parasite strains compared to directly sequenced samples. Conclusion: This optimized sWGA approach is a reliable method to obtain WGS data from non-leukocyte depleted, low parasitemia samples. The absence of amplification bias in data generated from mixtures of isolates from the same geographic region suggests that this approach can be appropriately used for molecular epidemiological studies.