Introduction

Human malaria, caused by protozoan parasites of the Plasmodium genus, is prevalent across many regions of the world, with the highest burden in Africa, followed by Asia, and then the Americas1. There are over 100 species of Plasmodium, but only Plasmodium falciparum, P. vivax, P. ovale curtisi, P. ovale wallikeri, and P. malariae are considered human malaria parasites. There are non-human malaria parasites that cause zoonotic infections in humans, among them P. knowlesi is known to infect humans residing in Southeast Asia2,3. Recent reports have identified several other zoonotic infections in humans4. Human malaria caused by P. brasilianum has also been reported from the Amazon basin region of South America5.

Plasmodium brasilianum is a malaria parasite of non-human primates found thus far in 13 genera and 36 species of New World monkeys across Central and South America (Fig. 1)6,7. Previous studies have shown strong similarities in the morphology and 18 s rRNA, mtDNA, and several surface protein gene sequences between P. brasilianum and P. malariae8,9. Furthermore, the amino acid sequence composition and protein structure are similar in the circumsporozoite protein (CSP) and merozoite surface protein 1 (MSP1) between the two species5,10.

Figure 1
figure 1

Global distribution of P. malariae and P. brasilianum. Reported distributions of P. malariae (red), P. brasilianum (blue), or both (purple) globally based on previous publications 6,7. While P. brasilianum has been reported in Central America in the past, there have been no recent reports of its presence or risk to humans outside of South America.

Full size image

These similarities have led to the suggestion that the two species are indeed the same species, with P. brasilianum originating from a spillover of P. malariae into nonhuman primates5. However, this hypothesis could not be thoroughly evaluated due to the lack of well-curated reference genomes for either species. The first complete genome for P. malariae highlighted the similarity between the two species7. To date, there has not been a whole reference genome available for P. brasilianum. A draft genome for P. brasilianum generated previously by our team from short-read sequencing revealed that the mitochondrial and apicoplast genomes of the two species were 99.7% and 99.6% identical between P. brasilianum and P. malariae, respectively11.

In this paper, we extend our previous effort using a combination of optical mapping, longer read PacBio sequencing, and Illumina short-read sequencing to generate the first complete reference genome of P. brasilianum. Using the previously published reference genome for the human infecting parasite P. malariae7, we compared the two genomes, which reveal strong genetic similarities. This is readily apparent in our comparisons of the msp3 and msp7 regions between the two species. The multigene members of each gene family, comprising 12 and 7 members, respectively, were organized in the same orientation in both genomes, with several nucleotide changes comprising the major differences observed.

Overall, we were not able to determine whether P. brasilianum and P. malariae are the same species from only a single genome from each species, but the availability of the P. brasilianum reference genome will facilitate the identification of species-specific diagnostic targets for detecting zoonotic transmission and will help drive future research related to Plasmodium evolution.

Results

Genome assembly and annotation

The host depleted QC passed reads from the two Illumina MiSeq libraries, and the PacBio RSII long reads were assembled using the hybrid assembly approach implemented in MaSuRCA12. The initial assembly consisted of 148 contigs. Circlator13 and IPA (https://github.com/ThomasDOtto/IPA) were used to merge overlapping contigs and generate scaffolds. ABACAS14 was used to port over chromosome names from equivalent contigs annotated in the P. malariae genome7. The assembled scaffolds were polished and SNPs and InDels with respect to P. malariae were identified using PILON15. The final assembly contained the 14 chromosomes, the mitochondrial genome, the apicoplast genome, and 29 unplaced contigs (Fig S1). To date, this is the most complete genome that has been generated for P. brasilianum. The assemblies of the 14 chromosomes were further optimized using optical mapping data; we observed near telomere-to-telomere assembly for 9 of 14 chromosomes when comparing the optical maps at the AfIII break-sites. We report the longest contiguous sequence for the remaining five chromosomes (PB01, PB02, PB03, PB04, PB06). The reported chromosomes show a high degree of similarity when compared with the P. malariae chromosomes. Pairwise sequence alignments between the two genomes show no structural differences (data not shown).

Genomic characteristics of Plasmodium brasilianum

The final assembled P. brasilianum genome is 31.4 Mb in length, consisting of 14 chromosomes, the mitochondrial genome, the apicoplast genome, and 29 unplaced contigs (Table 1). The contiguous genomes are 29.1 Mb in size and have 24.7% GC content, whereas the unplaced contigs all together are 2.29 Mb in size and have 20.6% GC content (Table S1). In contrast, the P. malariae genome7 is 33.6 Mb in length and has 47 unplaced contigs. The P. malariae contiguous genomes are 29.6 Mb in size and have 24.9% GC content, whereas the unplaced contigs all together are 4.04 Mb in size and have 18.9% GC content. If only the 14 mapped chromosomes and organellar genomes are considered, the two genomes vary in size by 0.461 Mb (1.57%) and GC content by 0.11%, making them 98.4% similar in size and 99.9% similar in GC content (Table S1).

Table 1 Assembly and annotation summary for P. brasilianum and P. malariae.
Full size table

The 14 mapped chromosomes and the organellar genomes were further evaluated using the LastZ aligner16 in the Geneious suite17. Out of the 29.1 Mb reference P. brasilianum genome, 88.4% was aligned between both genomes by the LastZ alignment and 98.4% of these nucleotides were identical (Table S1). Alignments done with Mauve18 demonstrated that most genomic differences between the two genomes were localized in the telomeric ends of chromosomes (data not shown).

As described in the methods, we used GeneMarkES19 and Augustus20, along with manual curation, to annotate 5924 genes, 144 of which belonged to the pir gene family, 213 to fam-l gene family, and 157 to the fam-m gene family. While structurally, we find that P. malariae and P. brasilianum share a high degree of similarity, we observe subtle differences when comparing the two species' overall gene content and identity (Fig. 2). A total of 5559 (93.8%) P. brasilianum genes were annotated across the 14 P. brasilianum chromosomes. By comparison, in P. malariae a total of 5,458 (83.0%) genes were annotated across the 14 P. malariae chromosomes. Throughout both genomes, a total of 4,970 (75.5%) P. malariae genes and pseudogenes had 80% or higher sequence similarity with 4,864 (82.1%) P. brasilianum genes.

Figure 2
figure 2

Genome organization of P. brasilianum and its variation from P. malariae. The four concentric rings, from outermost to innermost, show (1) the sizes and labels of the 14 P. brasilianum chromosomes, as well as the Apicoplast and Mitochondria; (2) the location of the 5,559 P. brasilianum genes (blue), excluding those on unplaced scaffolds; (3) the location of the 5,010 P. malariae genes (blue) and pseudogenes annotated onto the P. brasilianum genome through 80% or higher sequence similarity, excluding those on unplaced scaffolds; (4) line plot indicating SNP density of the 41,125 sequence variants between P. brasilianum and P. malariae sequences aligned by LASTZ alignment. The chromosome sizes and SNP density tracks are labeled with corresponding colors. The figure was generated using the Circa software.

Full size image

Evolutionary characteristics of Plasmodium brasilianum

To further understand the divergence between P. brasilianum and P. malariae, we identified 92 orthologous groups that have a crucial role in the invasion of erythrocytes21 (Table S2). There were 82 orthologous groups (1,567 genes) that have an ortholog in all currently available draft or complete Plasmodium genomes. An evolutionary tree was constructed with protein sequences from these genes using RaxML22 to better understand the evolutionary relationship of P. brasilianum with other Plasmodium species (Fig. 3A). The tree shows that P. malariae is the closest related species to P. brasilianum. The same result is seen using a subset of 90 orthologous groups (815 genes) that have an ortholog in the nine species previously considered 7 (Fig. 3B).

Figure 3
figure 3

Phylogenetic distances among Plasmodium species visualized by orthologous genes. Multiple sequence alignment was performed across Plasmodium species and an evolutionary tree was generated using the PROTGRAMMAAUTO model of RaxML. (A) Phylogenetic tree from a sequence alignment using 82 orthologous groups (1,567 genes) found in all 19 Plasmodium species with an available draft/complete genome. (B) Phylogenetic tree from a sequence alignment using 90 orthologous groups (815 genes) found in nine representative Plasmodium species7. To the right of each phylogenetic tree is a visual depiction of the primary vector each Plasmodium species infects during the parasite’s asexual life cycle. Colored boxes (blue in A and green in B) depict the locations of Plasmodium brasilianum and Plasmodium malariae.

Full size image

Comparative analysis between human malaria parasite Plasmodium malariae and Plasmodium brasilianum

From the genomic characteristics and evolutionary analysis described above, we observed that P. brasilianum and P. malariae share 97% pairwise identity as determined by LastZ alignments. Despite this, most recorded P. brasilianum infections have been isolated to new world monkeys, while P. malariae is predominantly found to infect humans in South America and Asia. To evaluate if there is sequence evidence for host specificity, we compared protein sequences from all available P. malariae and P. brasilianum accessions of 20 genes previously reported to affect host and vector invasion23 (Table 2). Of these, three genes had insufficient accessions to evaluate sequence variations, eight genes had minor or no sequence variation, and nine genes had sequence variations among accessions of the two species. Out of those nine genes, the csp gene is the most studied and was further evaluated.

Table 2 Comparative analysis of genes related to vertebrate host and vector invasion.
Full size table

The CSP protein plays a crucial role in the parasite’s invasion into the salivary glands of the vector24. Comparing our csp sequence along with all the published sequences for csp from P. brasilianum and P. malariae (Table S3), we observed variability in the large tetrapeptide repeat (NAAG/NDAG/NAPG/NDEG) contained within the gene, whereas the non-repeat regions of the gene were almost identical (Fig. 4A). Comparing the lengths between the two species, we see that the tetrapeptide repeat length in P. malariae from infected humans from Asia and South America lies in the range of 55–60 nucleotides, while the sequences from Africa show greater variability in repeat lengths (Fig. 4B). By contrast, csp sequences from non-human primate infected parasites, namely a Chimpanzee infected P. malariae and seven sequences from New World monkey infected P. brasilianum samples, had shorter repeat segments than human infected P. malariae and P. brasilianum samples. This presents an interesting finding, suggesting that CSP may play a role in host specificity and adaptation and provides a direction to pursue in future studies.

Figure 4
figure 4

Csp tetrapeptide repeat length variation among P. brasilianum and P. malariae samples. (A) Multiple Sequence Analysis of the circumsporozoite protein Csp annotations from NCBI for P. malariae and P. brasilianum. The boxed-out region indicates the location of a tetrapeptide repeat (NAAG/NDAG/NAPG/NDEG) with variable length among the samples. (B) Comparison of csp repeat lengths in nucleotides between P. malariae and P. brasilianum in samples from different continents and hosts. The P. brasilianum sample from this study and the P. malariae sample from Rutledge et al.7 are labeled with a blue and a red box, respectively.

Full size image

In addition to csp, we evaluated the genome organization and sequence similarities between the P. brasilianum and P. malariae merozoite surface protein (MSPs) genes. MSPs genes mediate the initial contact stage of erythrocyte invasion by the parasite25. Of particular interest are the multifamily gene members of MSP3 and MSP7, each of which are organized into a contiguous genomic region and have been pursued as potential vaccine candidates26,27. MSP1 and MSP9 gene sequences from P. malariae were also compared and found to be most similar to their P. brasilianum ortholog, but since they make up the sole members of their respective families they were not analyzed further. Each of the 12 msp3 genes (Fig. 5A, top) and the seven msp7 genes (Fig. 5B, top) from P. malariae were most similar to their P. brasilianum ortholog. Furthermore, the msp3 (Fig. 5B, bottom) and msp7 (Fig. 5B, bottom) gene regions were organized the same way in both genomes. This tendency for msp3 or msp7 paralogs to be organized in a clustered array on a chromosome is not surprising as it has been previously reported for both gene families in most Plasmodium species with a completely annotated genome28,29. The notable exception is that three P. malariae genes (PmUG01_12030100, PmUG01_12030200, PmUG01_12030300) aligned to a single P. brasilianum gene (gene4010). Closer examination revealed better overlap with these genes with the three different coding sequences of gene4010, pointing to differences in annotations for nearly identical sequences between the two genomes. Further work utilizing RNA-based sequencing experiments is needed to validate and optimize these annotations.

Figure 5
figure 5

Phylogenetic distances and genome organization of msp3 and msp7 regions. Multiple sequence alignment was performed using members of the merozoite surface protein (msp) genes and a phylogenetic tree was generated using the Bayesian method implemented in MrBayes v3.2.6. (A) Phylogenetic tree (top) from a sequence alignment using 12 P. malariae msp3 genes and their corresponding sequence matches in P. brasilianum. Branches from P. malariae genes are grey with their unique ID label shown in bold, and branches from P. brasilianum are green. Genome browsers snapshot (bottom) of the msp3 region from both species. Genes demonstrated in the phylogenetic tree are in bold. Consensus between the two genomes is shown at the top of the browser shot with differences depicted as dips in the green heatmap and as black marks in the grey tracks for each species. (B) Phylogenetic tree (top) and genome browser snapshot (bottom) from a multiple sequence alignment using 6 P. malariae msp7 genes and their corresponding sequence matches in P. brasilianum. (Pm = Plasmodium malariae, Pb = Plasmodium brasilianum; Green = P. brasilianum genes, orange = coding sequences, yellow = P. malariae protein coding genes, pink = P. malariae pseudogenes, blue (HP) = P. malariae hypothetical proteins; 5′ = 5′ end coding sequence, M = middle coding sequence, 3′ = 3′ end coding sequence).

Full size image

Discussion

Malaria remains a major public health issue affecting 85 countries with endemic disease contributing to an estimated 241 million cases and 627,000 deaths in 20201. Significant gains have been made in reducing malaria burden using traditional malaria control tools, although recently progress has slowed. Introduction of novel tools, including the introduction of the RTS,S/AS01 malaria vaccine that targets the circumsporozoite protein of P. falciparum, may help enhance efforts to reduce malaria transmission30. In addition to vector control tools, case management remains an important pillar for malaria control and elimination. As progress is being made in eliminating malaria in Southeast Asia, zoonotic hosts of malaria parasites like P. knowlesi have become large reservoirs of malaria and are challenging malaria elimination efforts31,32,33. Simple diagnostic tools such as microscopy and rapid diagnostic tests are not sufficiently sensitive to make accurate diagnosis of P. knowlesi; molecular tools are necessary for diagnostic confirmation. Even with powerful protocols for rapidly isolating DNA, creating purified libraries, and obtaining millions of next generation sequenced reads, the obtained sequences are not informative if the targeted sequence is the same in multiple species, as would be the case for P. malariae and P. brasilianum targets that are nearly identical, or if the targeted sequence is not known for species that may be present in the sample. In this context, the availability of a P. knowlesi genome sequence allowed for the identification of better diagnostic targets34.

Furthermore, zoonotic malaria cases caused by zoonotic Plasmodium species are expected to rise, increasing the need for accurate diagnosis. This is due to two main factors: (1) human contact with nonhuman primate Plasmodium reservoirs, facilitated by mosquito vectors, is expected to rise in areas of the world where deforestation increases the probability of zoonotic transmission; (2) surveillance and treatments focused on eliminating malaria caused by human-infecting Plasmodium species may create an opportunity for infections in humans by zoonotic Plasmodium species35. For example, Malaysia has reported zero indigenous non-zoonotic malaria cases in the last four years but has struggled with cases of zoonotic malaria due to P. knowlesi1. Experimentally, several zoonotic malaria parasites were able to be transmitted to humans by mosquito bites, including P. knowlesi (macaque monkeys of Southeast Asia)36,37, P. cynomolgi (macaque monkeys of Asia)38,39,40, P. brasilianum (multiple platyrrhine monkey species of South and Central America) 40, and P. inui (several monkey species of Asia)41. A naturally acquired human infection has been reported for each of these species since: P. knowlesi42, P. cynomolgi4,43, P. simium44, P. brasilianum5, and P. inui4,45.

Accurate detection of parasites in these zoonotic infections in humans is important for surveillance and relies on conserved primers, able to bind to genomes of multiple species, to amplify genes or genomic stretches of interest from extracted DNA for use in sequencing or restriction digestion-based identification assays. Deciding which targets to utilize for accurate identification can be greatly improved by the availability of complete reference genomes for these species. Reference genomes exist for P. knowlesi (strain A1-H.146, strain H47, and strain Malayan Strain Pk1 A48), P. cynomolgi (strain B49 and strain M50), P. inui (strain San Antonio 1, GCA_000524495.1), and P. simium51. The first complete genome for P. brasilianum reported here provides the toolkit needed for accurate detection and understanding of zoonotic malaria from these emerging sylvatic reservoirs52.

Rutledge and colleagues published a complete genome in 2017 for P. malariae whose 18S rRNA sequences were indistinguishable from P. brasilianum7. The similarity in the 18S gene sequences between the two species has been observed before. In addition, the similarity in csp sequences between the two species further emphasized the lack of genomic distinction53. Nonetheless, P. brasilianum and P. malariae have been maintained as distinct species since 1931. This was based on an experiment by Clark and Dunn, who demonstrated that seven patients given parasitized blood from a black spider monkey did not show evidence of successful infection54. The recent finding of a natural infection of P. brasilianum in humans disputes the notion that P. malariae and P. brasilianum are simply closely related and separated based on host specificity5. In fact, genomic comparisons support the hypothesis that P. brasilianum and P. malariae may be the same species5. Indeed, P. malariae has been found infecting African apes demonstrating the plasticity of this species in terms of the hosts it can infect55.

We observed 97.3% pairwise identity from LastZ alignments between our chromosomes and organellar genomes and those assembled for P. malariae by Rutledge et al.7 (Table S1). From these pairwise analyses, we identified a total of 41,125 non-synonymous SNPs between P. malariae and P. brasilianum (Table S4). These nucleotides constitute less than one percent of the genome. In fact, a total of 98.4% of the aligned regions between the two genomes were identical (Table S1). Many of the observed differences can be attributed to the disparity in the number of non-chromosomal contigs and the annotations for each genome. The P. brasilianum genome has 29 unplaced contigs, compared to 47 from P. malariae. Overall, the P. brasilianum genome has a higher resolution, with more telomeric sequences mapped. The telomere ends represent a challenging chromosomal region for assembly, and we utilized optical maps during contig assembly to assist with assembly at these regions56. We observed a total of 4,970 P. malariae genes that had 80% or higher sequence similarity with 4864 P. brasilianum genes (Table S5). Several P. brasilianum genes overlapped with more than one P. malariae gene. This can be seen in our msp7 comparisons (Fig. 5B), where a single P. brasilianum gene overlaps with three P. malariae genes. We cannot distinguish which annotation is more accurate without real expression data.

Gene annotations could be improved by incorporating RNA-seq expression data from P. brasilianum to annotate its transcriptome. Depending on the experimental design, a transcriptome typically only represents a snapshot of expression specific to a given cell type, developmental stage, or external stimulus. However, this can be addressed by utilizing single-cell RNA sequencing (scRNA-seq), as58. In addition to providing readouts for genes expressed only during specific stages of development, incorporating scRNA-seq would expand upon alternative transcript annotations. Otherwise, High-throughput Chromosome Conformation Capture (Hi-C) technologies could be utilized, alongside additional optical mapping experiments, to resolve the mapping of unplaced scaffolds48.

In conclusion, this work provides the first complete reference genome of P. brasilianum. While our findings support the hypothesis that P. malariae and P. brasilianum are strains of the same species, their evolutionary relationship cannot be fully determined without additional genome sequences from additional isolates. The notable exception to this was our observation that csp repeat sequences from non-human infecting P. malariae and P. brasilianum samples were shorter compared to those from human infections (Fig. 4B). Further studies are needed to evaluate the significance of this finding.

Methods

Sample origin and DNA isolation

Genomic DNA was extracted from ex vivo mature schizont stage parasites of the Bolivian I strain of P. brasilianum using the Qiagen DNA Blood Kit (QIAGEN, CA, USA, cat:51,104) and the MagAttract high-molecular-weight (HMW) DNA kit (Qiagen, MD, USA, cat: 67,563).

Sequencing

Genomic DNA libraries were prepared from isolated DNA using the NEBNext Ultra library prep kit (New England Biolabs, MA, USA, cat: E7370S) and the standard PacBio 20-kb procedure (Pacific Biosciences, CA, USA). Two separate runs were performed on the MiSeq using the MiSeq 500 cycle reagent kit (Illumina, CA, USA, cat:MS-102–2003). Two additional runs were performed on the PacBio RSII with C4 v2 chemistry for 360-min movies (Pacific Biosciences, CA, USA).

Quality filtering

FastQC57 was performed on the sequenced reads to analyze the quality of the Illumina and the Pacbio sequences. Raw reads were aligned against the host genome (GenBank accession no. 1GCA_000235385.1) and the reads that mapped to the host genome were discarded. Adapter trimming and quality control of the Illumina sequences were performed on the remaining reads using BBDuk from the BBMap toolkit58. Sequences with mean quality lower than 30, and length less than 50 were discarded from further analysis.

Genome assembly

The MaSuRCA hybrid genome assembler12, with default settings, was used to assemble the two Pacbio sequenced libraries and the Illumina paired-end sequenced libraries. Overlapping contigs from the MaSuRCA assembly were merged using Circlator13. Since the P. brasilianum genome and the mitogenome are linear genomes, with the apicoplast organized into a circular genome, Circlator was used to merge overlapping contigs into extended linear contigs59,60. The IPA script (https://github.com/ThomasDOtto/IPA) was run on the extended contigs to filter out contigs smaller than 5 kb. Contigs that are contained up to 90% within another contig were removed. Contigs with Illumina coverage greater than 50% of median coverage were merged. The resulting contigs were ordered and oriented with Abacas2 (https://github.com/satta/ABACAS2) by using the P. malariae genome as a reference. Finally, PILON15 was used for SNP calling and error-correcting in contigs. The apicoplast and mitochondrial genomes from the prior draft assembly11 were added to the extended, corrected assembly to generate the final complete assembly.

Genome annotation

Gene annotation was performed using both the de novo gene prediction tool GeneMark ES19, and Augustus20 that was trained on gene models from P. vivax. The resulting gene models were mapped against NCBI’s RefSeq Non-redundant protein database using the Diamond protein aligner61, in addition to manual curation. Orthologs across all 19 known Plasmodium species were identified using Orthofinder62. Blast2Go63 was used to perform GO enrichment analysis and Pfam annotations.

Annotation comparison

The Geneious Prime suite (Geneious Prime 2021.2.2. (https://www.geneious.com)) was used to visually compare the annotations between the P. brasilianum genome and the P. malariae genome from Rutledge et al.7. Each P. malariae chromosome that was structurally complete from end to end was aligned to its corresponding P. brasilianum chromosome one at a time using LastZ16,64, with default parameters and by using the P. brasilianum chromosome as the target sequence (Table S1). SNPs (Table S4) were identified from these alignments using Geneious’s “Find Variations/SNPs” option, with the options set to only find variants Inside CDS with a minimum coverage of 1. Synonymous mutations were excluded from the results. To visually place where each P. malariae gene annotation was aligned to the corresponding chromosome in P. brasilianum, the “Live Annotation and Predict” option in Geneious was used to transfer annotations from P. malariae chromosomes onto corresponding P. brasilianum chromosomes when the annotation’s sequence had at least 80% similarity. Overlapping genes were organized into a translation sheet for quick 1:1 comparison between the two genomes at these highly similar sequence regions (Table S5). A circa plot (Fig. 2) was created to demonstrate the size of each non-scaffold P. brasilianum chromosome, P. brasilianum gene annotations, transferred annotations from P. malariae, and SNP locations using the Circa software (Circa, available at http://omgenomics.com/circa).

Comparative genomics

We identified 92 orthologous groups that have at least one ortholog in most of the Plasmodium genomes that have a crucial role in the invasion of erythrocytes21 (Table S2). For all 19 currently available draft or complete genomes, there were 82 orthologous groups (1567 genes) that have an ortholog in all 19 genomes. When the genomes are restricted to only those species that were considered by Rutledge et al.7, namely P. vivax, P. brasilianum, P. berghei, P. falciparum, P. knowlesi, P. reichenowi, P. chabaudi, P. gallinaceum, and P. malariae, a total of 90 orthologous groups (815 genes) were found across in all nine genomes. Multiple sequence alignment using these 82 or 90 orthologous groups was performed using MUSCLE across all 19 or 9 Plasmodium species using Gblocks65. RaxML22 was used to generate evolutionary trees using the PROTGAMMAAUTO model to learn the likelihood model for the alignments.

To understand the difference in host interaction between the closely related species P. brasilianum and P. malariae, gene sequences from previously reported genes23 annotated as functionally important for host and vector invasion were compared among accession from P. brasilianum and P. malariae. The circumsporozoite adhesion protein coded for by the csp gene had multiple accessions for analyses for both species and had ample protein sequence variation among the accession and was studied further (Table S3). Multiple sequence alignment was performed using MUSCLE v3.8.42566, and the difference in the characteristic sequence repeats in the CSP protein was visualized with the Geneious Alignment View tab after using the Geneious Tree Builder with default parameters, as well as the Seaborn67 catplot function.

Additionally, several members of the merozoite surface protein gene families were compared, most notably msp3 and msp7. Gene sequences from P. malariae were aligned to corresponding gene chromosomes in P. brasilianum with the Geneious read mapper to identify the corresponding orthologs. The bedtools68 “getfasta” function was used to extract the sequence for each gene from both species (Table S6), and multiple sequence alignment was performed using ClustalX v2.0.1269 and MUSCLE as implemented in SeaView v4.3.570. Phylogenetic trees were generated using the Bayesian method implemented in MrBayes v3.2.6 with default parameters71, including a general time-reversible model with gamma-distributed substitution rates and a proportion of invariant sites (GTR + Γ + I). This was the best model that fit the data with the lowest Bayesian Information Criterion (BIC) scores, as estimated by MEGA v7.0.1472. Bayesian support was inferred for the nodes in MrBayes by sampling every 1,000 generations from two independent chains lasting 2 × 106 Markov Chain Monte Carlo (MCMC) steps. The chains were assumed to have converged once the potential scale reduction factor (PSRF) value was between 1.00 and 1.02, and the average SD of the posterior probability was < 0.01. Once convergence was reached as a “burn-in,” 25% of the samples were discarded.