Population genomics of Plasmodium ovale species in sub-Saharan Africa

Plasmodium ovale curtisi (Poc) and Plasmodium ovale wallikeri (Pow) are relapsing malaria parasites endemic to Africa and Asia that were previously thought to represent a single species. Amid increasing detection of ovale malaria in sub-Saharan Africa, we performed a population genomic study of both species across the continent. We conducted whole-genome sequencing of 25 isolates from Central and East Africa and analyzed them alongside 20 previously published African genomes. Isolates were predominantly monoclonal (43/45), with their genetic similarity aligning with geography. Pow showed lower average nucleotide diversity (1.8×10−4) across the genome compared to Poc (3.0×10−4) (p < 0.0001). Signatures of selective sweeps involving the dihydrofolate reductase gene were found in both species, as were signs of balancing selection at the merozoite surface protein 1 gene. Differences in the nucleotide diversity of Poc and Pow may reflect unique demographic history, even as similar selective forces facilitate their resilience to malaria control interventions.


Introduction
4][5] Over the last few decades, genomic studies of P. falciparum have enabled monitoring of drug resistance markers, 6 facilitated the identification of promising vaccine candidates, 7 uncovered the structure of parasite populations, 8 and identified evolutionary forces shaping their demography. 9,10Much less is known about nonfalciparum species, especially their comparative evolutionary history and susceptibility to malaria control interventions focused on P. falciparum.Plasmodium ovale was first identified as a separate malaria species in 1922 based on the appearance of oval-shaped erythrocytes that contained non-ring parasite forms. 11Hallmarks of this parasite species are its restriction to younger red cells and therefore propensity to cause low-density infections, as well as relapses from liver hypnozoites, similar to P. vivax and P. cynomologi.The species often causes co-infection alongside P. falciparum which, along with its low parasite densities, makes it challenging to differentiate morphologically on peripheral blood smears. 12The advent of polymerase chain reaction (PCR)-based diagnostics has improved detection of ovale infections, but initial PCR surveys across Africa and Asia based on the small subunit rRNA gene revealed two apparent groups of P. ovale parasites, termed classic and variant. 13The discovery of perfect sequence segregation of 6 genomic markers, and more recently 12 mitochondrial loci, between classic and variant P. ovale isolates collected across Africa and Asia has led to the conclusion that this dimorphism actually represents a true species divide in the P. ovale clade. 14,15 P. ovale curtisi and P. ovale wallikeri have since been confirmed to circulate within the same human populations throughout Africa and Asia, 19,20 with both detected by PCR at higher rates than previously appreciated. 5,12,21,22Limited investigation of the genetic diversity and population genetics of the two P. ovale species have hinted at low diversity and/or small effective population size, as few unique haplotypes have been identified at antigenic gene targets like apical membrane antigen 1 (ama1) and merozoite surface protein 1 (msp1). 20,23There is some indication that drugs used to treat P. falciparum are also shaping P. ovale parasite populations; signs of a selective sweep involving a mutant dhfr allele (implicated in pyrimethamine resistance) have been detected in both Poc and Pow. 24,25Until now, the low density of most P. ovale isolates combined with lack of an in vitro culture system has hindered whole-genome sequencing of these parasites. 1128] We employed hybrid capture or leukodepletion to enrich P. ovale spp.DNA and perform wholegenome sequencing (WGS) of 25 clinical isolates collected from studies conducted across Ethiopia, the Democratic Republic of the Congo, Tanzania, and Cameroon.Combined with 20 additional public whole genomes from 11 countries spanning East, Central, and West Africa, we sought to better understand the comparative biology of P. ovale curtisi, P. ovale wallikeri, and co-endemic P. falciparum by examining their complexity of infection, population structure, nucleotide diversity, and genomic signatures of selection.

High-quality genomic coverage of African P. ovale isolates
0][31][32][33][34] These included 13 P. ovale curtisi and 12 P. ovale wallikeri isolates that were selected from six studies based on robust amplification of the po18S rRNA gene (Ct <36) and predominance of one ovale species within each isolate.The majority of the samples (n=21) underwent a custom-designed hybrid capture with RNA baits to preferentially isolate ovale DNA extracted from dried blood spots for sequencing, while four additional whole blood samples were leukodepleted (LDB) at the time of collection by CF11 filtration and directly sequenced without enrichment. 35Finally, genomic data of 20 P. ovale isolates sequenced as part of four previously-published studies were retrieved from the European Nucleotide Archive and the Sequence Read Archive. 15,28,36,37These isolates either underwent selective whole-genome amplification (sWGA) or leukodepletion for parasite DNA enrichment prior to sequencing.Further data on all parasite isolates are found in Supplemental Table 1.Whole-genome sequencing achieved high genome coverage, with an overall average of 86% and 87% ten-fold coverage across the core genome for the 21 P. ovale curtisi and 24 P. ovale wallikeri isolates, respectively (Figure 1).Coverage and mapping proportion were highest when aligned to the P. ovale reference genome determined by the Poc/Pow species-specific qPCR assay, corroborating initial species assignment.Compared to sWGA and LDB samples, the hybrid capture method used to enrich parasite DNA in the majority of samples yielded more complete coverage across all chromosomes except for chromosome 10 (Figure 1).The hybrid capture was originally designed for P. ovale wallikeri, with additional P. ovale curtisi baits then selected to cover areas that differ between the two ovale genome assemblies (PowCR01 and PocGH01). 28Due to Pow chromosome 10 being incomplete in the PowCR01 reference genome (only 470kb), this approach did not provide coverage for the full Poc chromosome 10 (1,300kb).

Year of
This led to substantially lower coverage for chromosome 10 across all Poc hybrid capture isolates (40-60% 10x coverage vs. >85% for all other Poc chromosomes); thus, chromosome 10 was excluded from all genome-wide analysis in Poc isolates to limit error.As, expected, hybrid capture led to preferential sequencing of P. ovale DNA among samples that were co-infected with P. falciparum (Pf); Pf-positive isolates that underwent hybrid capture yielded only 2-11% 10x coverage of the Pf genome compared to >90% 10x coverage of the Pf genome among leukodepleted blood samples.For genomic analysis, insertions/deletions, multiallelic sites, low-quality variants, and variants within tandem repeats and expanded gene families were excluded (see Methods), yielding final biallelic single nucleotide polymorphism (SNP) call sets of 73,015 SNPs for P. ovale curtisi and 45,669 for P. ovale wallikeri.

Low complexity of infection
Complexity of infection (COI), or the number of unique parasite clones present in a given isolate, was estimated 1000 times using THEREALMcCOIL for all 21 P. ovale curtisi and 24 ovale wallikeri isolates, as well as 2,077 geographically matched P. falciparum isolates downloaded from the publicly available MalariaGEN Pf6 dataset (Figure 2). 38Twenty out of 21 Poc isolates (95%) and 23 out of 24 Pow isolates (96%) were estimated to be monoclonal; the remaining isolate in each P. ovale species was found to comprise two parasite clones.By comparison, roughly half (1,165/2,077; 56%) of P. falciparum samples were monoclonal.COI differed significantly (p = 0.005) among the three Plasmodium species.In pairwise comparisons, both Poc and Pow had significantly lower COI compared to P. falciparum (p = 0.001 and p=0.004, respectively).The two multiclonal P. ovale infections were both hybrid-capture-enriched, high coverage (94% and 98% 10-fold coverage in Poc and Pow, respectively) and came from high-transmission areas of the DRC; 39 each had a COI of 2. In order to determine whether the two clones in these samples were distinct lineages or meiotic siblings, we analyzed the distribution of heterozygous SNPs across the genome.We hypothesized that meiotic siblings would only have heterozygous SNPs in specific regions, reflecting recombination within the mosquito midgut. 40In both samples, after filtering to high confidence SNPs based on population-wide allele frequency, we saw an even distribution of heterozygous SNPs across the genome, suggestive of two distinct parasite lineages in the same host rather than meiotic siblings (Supplemental Figure 1).

Lower nucleotide diversity in P. ovale wallikeri compared to P. ovale curtisi
Among a collection of 3,339 sets of one-to-one-to-one orthologous genes between the Poc, Pow, and P. falciparum genomes, we identified 2,008 sets that achieved high-quality sequencing coverage and had no overlap with masked genomic regions in any of the three

Parasite genomic similarity recapitulates geographic relationships
Genome-wide principal component (PC) analysis of the monoclonal samples of each ovale species revealed spatial arrangement of related parasites along PC1 and PC2 that aligns with their location of origin (Figure 4).These components accounted for 15.0% and 17.5% of the genetic differentiation in Poc and Pow, respectively.While cluster analysis by ADMIXTURE found the best fit when modeling each isolate as a separate cluster, except for one pair of isolates per species originating from the capital of Kinshasa in the DRC (in Poc) and the Amhara region in Ethiopia (in Pow), geographic alignment was evident in the PCA.For Pow, PC1 and PC2 appear to reflect an East-West axis and North-South axis, respectively, with samples from Ethiopia and South Sudan in the west divided from others by PC1.In the PCA for Poc, Ethiopian parasites also organized separately from other samples, as did isolates from Kinshasa in the DRC.The remaining Poc samples show some division between East, Central, and West Africa, though the alignment with geography is less consistent than in Pow.In both P. ovale spp., PC3 and PC4 further separated samples from various countries (Supplemental Figure 2).Examination of the top 0.5% of variants by contribution to each of the first 4 principal components revealed that SNPs within genes encoding multidrug resistance protein 1 (mdr1) and dihydrofolate reductase -thymidylate synthase (dhfr-ts), two putative antimalarial resistance genes, were major contributors to the North-South axis in Pow PC2.Among all 24 Pow samples, three previously-documented haplotypes in Pow dhfr-ts, a key gene in folate metabolism that is implicated in pyrimethamine resistance, 41,42 appear to drive this geographic differentiation, with the Phe57Leu+Ser58Arg haplotype existing in in 45% of our Central African clones and 36% of our East African clones but none of the sequenced West African clones (Supplemental Figure 3A). 25This haplotype is associated with resistance to pyrimethamine when expressed in E. coli.Though it did not drive differentiation in the PCA, Poc dhfr-ts haplotypes similarly showed presence of a putative drug resistance haplotype (Ala15Ser+Ser58Arg) in the Central and East African clones but not in West Africa, though our

Signatures of selection contain putative drug resistance loci, proteins involved in sexual stage differentiation, and antigenic targets
We calculated nS L and Tajima's D across the genomes to identify loci under directional and balancing selection, respectively.The nS L statistic is considered robust to the currently-unknown recombination rates across the genomes of P. ovale species. 43Genetic markers of interest within 10kb of the top 0.5% absolute normalized nS L values that may be influenced by selective sweeps are listed in Table 2 and Table 3.Evidence of a selective sweep involving the putative bifunctional dihydrofolate reductase -thymidylate synthase (dhfr-ts) gene 41,42 was found in both P. ovale species (Figure 5).Examination of extended haplotype homozygosity (EHH) at the selected variants show a large selective sweep in Pow spanning roughly 40kb as well as close proximity of the dhfr-ts gene to the focal variant (Figure 6A, 6B).In Poc, the positioning of the dhfr-ts gene lies at the edge of a smaller sweep (Figure 6C, 6D).However, another putative marker of drug resistance, multi-drug resistance associated protein 2 (mrp2), was found in close proximity to one of the highest absolute normalized nS L value in Poc and lies near the center of a 40kb sweep region on Poc chromosome 14 (Figure 6E, 6F).
5][46] In Pow, four top nS L hits were found around genes encoding the dynein heavy and light chains, cytoskeleton components highly expressed in male gametes for motility and fusion with female gametes in the mosquito blood meal. 47Top nS L hits were also found near cysteine-repeat modular proteins 2 and 1 (crmp2/1) in Poc and Pow, respectively, proteins which may be involved in targeting sporozoites to the salivary glands in the mosquito prior to transmission. 48,49ally, genes encoding putative antigenic targets at the host-parasite interface, including merozoite surface protein 7 (msp7), merozoite surface protein 5 (msp5), early transcribed membrane protein (etramp), apical membrane antigen 1 (ama1), GPI-anchored micronemal antigen (gama), and 6-cysteine protein B9 (6-cys) may be under directional selection in both P. ovale species. 50,51An orthologue of sporozoite protein essential for cell traversal 1 (spect-1), a protein necessary for liver cell invasion that has been investigated as a potential vaccine target, 52 was also among the top hits in Pow.
Overall, Tajima's D in both species exhibited a negative skew across the genome with an average value of -1.06 for Poc and -0.78 for Pow.This may suggest population expansion following a bottleneck or weak directional selection (Figure 7).Among the loci with positive values in the top 0.5% of absolute Tajima's D hits, the antigenic marker merozoite surface protein 1 (MSP1) was identified as a probable target of balancing or diversifying selection in both P. ovale species (Tables 2, 3).

Discussion
We present a comprehensive population genomic study of both P. ovale species within sub-Saharan Africa.Our study comprises 21 Poc and 24 Pow isolates selected from 11 studies, including both febrile and asymptomatic cases.Genome-wide analysis reveals differences in nucleotide diversity between P. ovale species, but similarity in their low complexity of infection, geographic relatedness, and signatures of selection.Our analysis was performed using genomic enrichment methods specifically designed to enable robust coverage and analysis with the 2017 reference genomes of P. ovale curtisi (from Ghana) and P. ovale wallikeri (from Cameroon), which were the available references at the time of the study. 28pared to the "classic" P. ovale curtisi species, we observed significantly lower nucleotide diversity across orthologous genes among geographically-matched P. ovale wallikeri isolates (2.5x10 -4 for Poc and 1.8x10 -4 for Pow, respectively).Our estimate for Poc is concordant with the genome-wide diversity calculated among six Central African Poc isolates 15 as well as that derived from RNA expression data among four parasite samples from Mali. 53wever, our Pow estimate was substantially lower than the genome-wide estimate reported by Higgins et al. (3.4x10 -4 ), despite our inclusion of their samples alongside additional Central and East African isolates.Our lower estimate may reflect the exclusion of higher-diversity intergenic regions, though we also found lower genome-wide and intergenic SNP density in Pow compared to Poc.Relatively low nucleotide diversity in Pow may indicate reduced effective population size, increased inbreeding, or a population bottleneck in the time since Poc and Pow diverged between 1.3 and 20.3 million years ago. 14,28More recent population growth in both species is also suggested by the predominantly negative distribution of Tajima's D values across their protein-coding genes, a finding that can indicate population expansion following a bottleneck. 54The high ratio of nonsynonymous-to-synonymous substitutions among these protein coding genes (2.5 for Pow, 1.5 for Poc) is similar to that seen in P. falciparum and P. vivax. 55,56This finding may represent diversifying selection on proteins across either P. ovale genome, enabling maintenance of nonsynonymous substitutions, or inflation of dN/dS ratios observed among Plasmodium parasites due to the impact of the malaria life cycle on allele frequencies. 57Further analysis of subpopulations of each parasite species could help to elucidate the factors driving the observed difference in genomic diversity, such as by determining whether Pow isolates from Asia have similarly low nucleotide diversity or if this finding is specific to Africa.
The observed predominance of monoclonal isolates among both P. ovale species is consistent with low within-sample haplotypic diversity seen in previous investigations of African P. ovale isolates by genome-wide RNA sequencing and amplicon sequencing. 53,58Low complexity of P. ovale infections may result from efficient clonal transmission 32 and/or lower transmission overall, limiting vector uptake of multiple parasite clones from either the same or different infected individuals.This low complexity is expected to limit opportunities for genetic recombination within mosquito vectors, though multiple-clone infections were identified in Kinshasa, a region with overall higher malaria endemicity and transmission intensity. 39c signatures of selection within both P. ovale species highlighted the importance of antimalarials, host-vector life cycle transitions, and human immunity as evolutionary pressures impacting parasite survival.P. ovale infections are frequently subclinical and go untreated, 59 but likely still face substantial drug exposure from widely used antimalarials prescribed for P. falciparum. 22Additionally, malaria prophylaxis using sulfadoxinepyrimethamine, such as intermittent preventive therapy in pregnancy (IPTp) and seasonal malaria chemoprevention for infants and schoolchildren (SMC), may be applying drug pressure on P. ovale parasite populations. 1 Selective sweeps in dhfr-ts, a gene implicated in resistance to pyrimethamine, have been documented in both Poc and Pow, and certain mutant alleles were found to confer pyrimethamine resistance when expressed in E. coli. 24,25In our dataset, sweeps near the dhfr-ts genes were among the strongest signals of directional selection in both P. ovale species, especially in Pow, possibly representing drug pressure influencing parasite survival.We also found the putative pyrimethamine resistance Pow dhfr-ts haplotype Phe57Leu+Ser58Arg to be a major contributor to principal component 2 of Pow (representing the North-South axis); the resistant haplotype composed 36% (9/25) of our Pow haplotypes, with representation in Central and East Africa but no detection in West Africa.Another putative pyrimethamine resistance haplotype (Ala15Ser+Ser58Arg) in Poc was similarly detected in East and Central Africa, though not in West Africa. 25Functional evaluation of different alleles to determine their capacity to confer drug resistance, as well as monitoring of these alleles across the parasite populations over time, will further clarify how interventions targeting P. falciparum may be simultaneously rendering P. ovale parasites harder to control.
Finally, strong signals of balancing or diversifying selection were observed in both species within their genes encoding merozoite surface protein 1 (MSP1), the orthologues of the predominant antigen on blood-stage P. falciparum parasites that has been shown to induce protective immunity in some studies. 60Diversifying selection on MSP1 has been documented in P. falciparum 61 as well as in focused analysis among African P. ovale infections imported to China. 62Our dataset provides even stronger evidence for diversifying selection at this site, as the msp1 gene showed the single highest Tajima's D value across all genes in both P. ovale species.Such immune responses may also play a role in modulating relapse potential. 63s study has several limitations.While it represents the largest genome-wide examination of the genomic composition of both P. ovale curtisi and P. ovale wallikeri to date, the sample size for both species is nonetheless small and limits the power to detect clustering of isolates, infer population demography, and detect selection.The geographic coverage of the isolates employed differs between the 13 countries represented, and isolates from Northern or Southern Africa were not available.Whole-genome enrichment methods also differed among isolates; 21 isolates employed hybrid capture, 15 used selective whole-genome amplification, and nine relied on leukodepletion.The two former methods may have induced amplification bias, whereas leukodepletion does not amplify P. ovale DNA and therefore may reduce the power to identify rare variants in those isolates.Disparate average read depth between the three species (44.1, 95.7, and 147.5 for Poc, Pow, and Pf, respectively) may also have differentially impacted our ability to detect polyclonal infections, but the low complexity of infection found in both P. ovale species should be robust given the satisfactory sequencing depth overall.The source studies also differ by whether samples were derived from asymptomatic carriers (n=11) or febrile patients (n=34).Sample sizes were too small to analyze these populations separately.Finally, hybrid capture baits designed using the incomplete PowCR01 reference genome led to incomplete coverage of Poc chromosome 10, which was excluded from analyses.Unfortunately, newly-assembled regions of Pow chromosome 10 were not available during the analysis.The hybrid capture approach also did not enable enrichment and analysis of loci in the mitochondrial and apicoplast genomes, which were excluded from analysis.We do not expect these exclusions to systematically bias estimation of nucleotide diversity nor complexity of infection, though it does prevent us from evaluating excluded loci (including mdr1, msp3, and msp8) for genomic signatures of selection.The availability of selective whole genome amplification protocols now provides a less expensive approach to targeted DNA enrichment for P. ovale spp.that does not rely on the specific design of hybrid capture baits. 27s study provides a comparative genomic analysis of the two Plasmodium ovale species sympatrically circulating in sub-Saharan Africa and presents new evidence of selective pressures on genes related to drug response, sexual differentiation, and immune evasion.Further population genomic studies of Poc and Pow should employ a larger selection of isolates from a greater geographic range, especially including Asia, and take advantage of new reference genome assemblies to build on these insights. 15Functional investigation into the genes showing signatures of selection, including via orthologue replacement in closely related Plasmodium species, 64,65 is also an exciting new strategy in substantiating the biological relevance of key loci, with implications for transmission prevention, treatment strategies, and vaccine development for P. ovale spp.Finally, cataloging genome-wide diversity facilitates the design of targeted genotyping methods that can efficiently characterize the epidemiology of these understudied parasite species. 66Combining these approaches to better evaluate P. ovale parasite relatedness, transmission, and relapse patterns can help to improve the impact of current malaria control strategies on all human-infecting malaria species. 36,67,68

Sample selection
Clinical isolates in the form of dried blood spots or leukodepleted blood were drawn from six studies shown in Table 1, including studies involving both asymptomatic persons and febrile patients across four countries.Across these studies, participants were screened for the presence of P. ovale spp.infection by a real-time polymerase chain reaction (qPCR) assay targeting the po18S rRNA gene. 22Among 282 isolates with a po18S Ct value under 40, a species-specific (Poc and Pow) 18S rRNA qPCR assay was employed to determine the ovale species present. 69Candidates were selected from isolates with only one species detected or mixed infections in which one species predominated by ≥3 Ct (corresponding to approximately 8 times as much DNA).Samples were also screened for presence of P. falciparum co-infection using a qPCR assay for the pfldh or pf18S rRNA gene. 31,39Ultimately, samples from 25 individuals were selected for whole-genome sequencing based on higher-density P. ovale infection, lack of or lower-density P. falciparum coinfection, and balance of ovale species and geographic diversity across the sample set.Characteristics of these 25 samples, and an additional 20 samples from four previously-published studies, 15,28,36,37 are shown in Supplemental Table 1.
Variant calling from aligned reads across each ovale species genome was also performed using the GATK best practices pipeline. 74In the resulting callset, variants were masked if they fell outside of the 14 chromosomes of the reference genome or were part of specific expanded gene families (see Supplemental Methods).GATK hard filtering was then used to remove variants with poor quality metrics using the following filter thresholds: quality by depth <3, Fisher strand bias >50, strand odds ratio >3, mapping quality >50, mapping quality rank sum <-2.5, read position rank sum <-3.Callsets were limited to biallelic single nucleotide polymorphisms (SNPs) that were present in at least 80% of individuals.SNP density across the entire genome and within specific functional regions of each genome were calculated using custom scripts (see Data Availability).SNPeff v4.3 was used to annotate individual variants and determine the ratio of nonsynonymous-to-synonymous mutations. 28,77

Selection of P. falciparum comparison dataset
Co-endemic P. falciparum samples were drawn from the Pf6 dataset. 38Of 20,705 total P. falciparum isolates from around the globe, 2,077 came from the same or nearby geographic locations as the source studies of P. ovale isolates described above.Thirty-two P. falciparum samples with over 85% base callability were randomly selected in order to have at least one coendemic P. falciparum isolate for each P. ovale curtisi or ovale wallikeri isolate (Supplemental Table 2).Variant callsets for these P. falciparum samples were limited to the falciparum core genome, 73 quality filtered by Variant Quality Score Recalibration, 74 and restricted to sites present in at least 80% of individuals.

Population genetic analyses
Complexity of infection was estimated for all samples using THE REAL McCOIL via the McCOILR package. 78Principal components were calculated and overall genomic similarity compared among monoclonal isolates using PLINK v1.90b6.21. 79Nucleotide diversity was estimated among sets of one-to-one orthologous genes between monoclonal Poc, Pow, and P. falciparum isolates using vcftools v0.1.15. 80For signatures of selection, nS L was calculated for all non-missing variants in monoclonal isolates of each P. ovale species using selscan v1.2.0, 81 while Tajima's D was calculated in 300bp sliding windows in protein-coding genes using vcf-kit v0.2.9. 82Additional details on statistical methodology can be found in the Supplemental Methods.

Table 1 . Studies of origin for 45 P. ovale isolates. *Raw sequencing data for these 20 isolates were directly incorporated into the analysis pipeline after retrieval from the European Nucleotide Archive or NCBI Sequence Read Archive.
species.The average species-specific nucleotide diversity ( ) among these orthologues in the 20 monoclonal Poc, 23 monoclonal Pow, and 19 geographically-matched monoclonal Pf samples were 2.9x10 -4 , 1.8x10 -4 , and 2.6x10 -4 , respectively.These were significantly different between species (p < 0.0001, F =98, df = 2), with orthologues in P. ovale curtisi more diverse than in P. ovale wallikeri and P. falciparum (p-values <0.0001 and 0.002, respectively), and Pow orthologues less diverse than in Pf (p < 0.0001) (Figure3A).To mitigate bias by geographic coverage and orthology with the P. falciparum genome, we repeated this analysis using 2,911 Poc-Pow orthologues among a group of geographically-matched monoclonal Poc and Pow samples (n=11 each, Supplemental Table3), revealing average nucleotide diversities of 2.5x10 -4 and 1.8x10-4, respectively (Figure3B).Nucleotide diversity was still significantly lower in Pow orthologues compared to Poc (p < 0.0001).This high nucleotide diversity in P. ovale curtisi was consistent with investigation of the total number and density of genome-wide SNPs.Variant calling and filtering resulted in almost twice the number of SNPs for P. ovale curtisi (73,015) as for P. ovale wallikeri(45,669), despite a slightly smaller number of Poc isolates(21 vs. 24).This corresponded to a higher density of SNPs across the Poc genome (4.0 SNPs per kilobase[kb] in Poc vs. 2.4 SNPs/kb in Pow).However, a smaller proportion of SNPs in the Poc genome were nonsynonymous mutations, as the ratio of nonsynonymous-to-synonymous (dN/dS) mutations was 1.5 and 2.5 in Poc and Pow, respectively.Among the aforementioned geographically-matched P. ovale isolates, dN/dS within 2,911 orthologous genes was 1.4 for Poc and 2.5 for Pow, consistent with the broader estimates.SNP densities in both the Poc and Pow genomes were lowest in protein-coding sequences (3.9 and 1.9 SNPs/kb, respectively), and higher in introns (4.4 and 2.8 SNPs/kb) (Figure3C).Intergenic regions in Poc showed relatively lower SNP density similar to proteincoding sequences (4.0 SNPs/kb), but these same regions in Pow had relatively high SNP density similar to introns (2.8 SNPs/kb).