Primers were designed to amplify a region of the 18S rRNA gene approximately 200 base pairs in length that is highly conserved across eukaryotic organisms yet contains sufficient diversity at the nucleotide level to allow accurate species identification and differentiation. The selected amplification region possesses BamHI and XmaI restriction enzyme cut sites only in the human host sequence (Additional file 1: Figure S1), which allows cleavage of host template and reduced amplification of host DNA during PCR amplicon enrichment prior to Illumina amplicon deep sequencing (Fig. 1). Following sequencing, paired and trimmed reads were mapped to a database of human and parasite 18S rRNA sequences, and the number of mapped reads per parasite species was counted. Given the sensitivity of Illumina sequencing, the occurrence of Illumina index cross-talk, and because some DNA cross-contamination can be expected in samples that are extracted and processed together [23–25], we established a conservative, dual-criterion system to differentiate “noise” from a true positive sequencing result. This system utilizes a minimum cutoff for positivity based on the average proportion of contaminating reads obtained per negative control specimen over multiple replicate analyses (assuming 60–80 samples are multiplexed in a single library, see the “Methods” section for further details). In addition, a species-specific shifting maximum cutoff value was established to be applied on a per specimen basis and to account for minor changes in the degree of index cross-talk and variations in the number of reads generated per specimen/experiment. Using this dual-criterion system, specimens were considered positive only if more than 20 reads mapped to the respective parasite reference sequence and if the number of parasite derived reads mapped to that reference sequence also exceeded the shifting maximum cutoff value.Fig. 1

Reduction of host DNA by restriction enzyme digestion enhances PCR amplification of parasite DNA. DNA extraction from parasite-infected whole blood yields a DNA sample containing high amounts of host DNA (blue) and low amounts of parasite DNA (bright red). a Performing conventional PCR on this sample, using universal primers, amplifies primarily host DNA (blue), and yields sequencing reads almost entirely belonging to the host. b In contrast, restriction enzyme digestion of host DNA prior to PCR alters the ratio of host to parasite DNA in the initial sample, allowing for selective amplification of parasite DNA (bright red) and resulting in an increase in the relative number of parasite amplicons post-PCR and an increase in the sensitivity of parasite detection via NGS

Human or (surrogate) animal blood samples containing previously diagnosed parasitic infections (Table 1) were processed in triplicate, as shown in Fig. 1b. Bioinformatic analysis indicated that the primer binding sites, the region of amplification, and the restriction enzyme cut sites of all relevant animals share 100% identity with the human target sequence. Consequently, samples containing animal blood were analyzed in an identical fashion to human clinical samples. Most blood-borne parasites known to infect humans were included in our analysis; however, we were unable to obtain clinical samples/isolates for Trypanosoma brucei subsp. gambiense and several rare human filarial blood parasites. Bioinformatic analysis of the complete 18S rRNA sequence from several blood parasites identified BamHI and/or XmaI cut sites outside the region of amplification, varying slightly in their location and frequency between parasite taxa. As such, the resulting restriction fragments would vary in size for different blood parasites. Consequently, post-digestion DNA for all validations was divided into two equal parts and cleaned using a Monarch PCR & DNA Cleanup Kit selecting for both > 2 kb and < 2 kb DNA products, respectively. Prior to DNA extraction and restriction digestion, all samples were spiked with cat blood containing 3.4 × 10⁶ Cytauxzoon felis parasites as an extraction, amplification, and sequencing internal control.Table 1

Host, source, original identification method and GenBank accession number of samples used in this study

Following TADS, a substantial reduction in reads mapping to the human host reference was observed in the digested samples compared to undigested samples (Fig. 2a). Furthermore, the digested samples showed a 5- to 10- fold increase in the number of parasite-specific reads (Fig. 2a and Additional file 2: Figure S2). All samples assayed passed our set criteria for positivity except for two Wuchereria bancrofti blood samples. For these samples, the failure to detect W. bancrofti was attributed to the formation of a large, non-uniform clot that made DNA extraction problematic (Additional file 3: Figure S3a-c). Nevertheless, these data suggest that reduction of host DNA background via restriction enzyme digestion improves detectability of parasite DNA for the universal detection of parasites in blood. Analysis of post-digestion size selection found no statistical difference between > 2 kb and < 2 kb cleanup conditions (p = 0.0631).Fig. 2

Digestion of host DNA increases the sensitivity of parasite detection in parasite-positive human blood samples. (a) Restriction enzyme digestion yields a marked reduction in human 18S rRNA reads per thousand (left panel, greyscale diamonds) and a 5- to 10-fold increase in parasite reads per thousand (right panel, colored circles) in digested relative to undigested samples (n = 3 biological replicates, mean ± SD, samples were normalized according to the reads per thousand for reads derived from human host and parasite separately, with the central dotted line reflective of a zero fold change, which marks the undigested samples before treatment with restriction enzymes). No statistical difference was found for size selection (i.e., > 2 kb vs. < 2 kb) (two-way ANOVA, p = 0.0631). (b) Proportional composition of human DNA dilutions in undigested (ud) and digested (d) samples demonstrates an average 2-fold reduction in human DNA and a 5-fold increase in parasite reads post-digestion (black bars = C. felis, dark grey bars = H. sapiens, light grey bars = P. falciparum, concentration of 3D7 DNA includes P. falciparum and H. sapiens DNA from 3D7 cultures which contain human blood products, two-way ANOVA with Sidak’s multiple comparisons posttest, p < 0.0001, n = 3, mean ± SD)

To determine the extent to which enzyme digestion reduces human host background, a series of dilutions was prepared using human DNA donated by healthy volunteers and parasite DNA obtained from a 3D7 P. falciparum culture. Samples contained 0.2 ng/μL P. falciparum DNA, or approximately 8600 parasites per microliter, as well as human DNA diluted to a concentration of 3, 2.5, 2, 1.5, 1, 0.5, or 0 ng/μL. Bioinformatic analysis indicated no BamHI/XmaI cut sites within 2 kbp of the PCR amplification region for P. falciparum 3D7, so all post-digestion samples were cleaned according to the > 2 kb size selection protocol. In undigested samples, 80–100% of sequencing reads mapped to the human host reference sequence with only 0–20% of reads mapping to P. falciparum (Fig. 2b). Meanwhile, the composition of digested sample reads was only 40–50% human and 50–60% P. falciparum, reflecting a greater than or equal to 2-fold decrease in human reads and an average 5-fold increase in parasite reads post-digestion (Fig. 2b).

As an additional control, a mock restriction digestion was performed in triplicate wherein DNA extracted from human blood spiked with cultured 3D7 parasites was incubated for the same period and at the same temperature as identical specimens subjected to a true restriction enzyme digestion. These samples were subsequently purified using a Monarch Cleanup Kit (> 2 kb) and PCR amplified according to the protocol described. Post-sequencing analysis found no statistical difference in the number of 3D7-derived sequencing reads obtained between mock digested DNA specimens and their paired undigested DNA samples, confirming that the increase in parasite reads in the restriction-digested samples is directly related to the action of the restriction enzymes on host DNA (Additional file 4: Figure S4).

To explore the effectiveness of this method for detecting mixed infections, a variety of mixed parasite blood samples were artificially produced by combining previously diagnosed parasite-infected blood samples and subjecting them to analysis via the described method. The samples simulated all varieties of mixed malaria infections, including all the major Plasmodium species that infect humans, together and in pairs, as well as other geographically possible mixed infections. As before, we saw a dramatic decrease in human reads in digested relative to undigested samples and, in this case, a 2- to 15-fold increase in parasite reads post-digestion (Fig. 3, left panel). These data establish that this method can reliably detect parasites in mixed infections but suggest that competitive amplification occurs between the different 18S rRNA types, which may affect the sensitivity of detection for parasite species occurring at relatively low numbers in mixed parasite communities.Fig. 3

Enzyme digestion enhances sensitivity of detection for mixed parasite infections. Restriction enzyme digestion of mixed parasite infections in human blood yields a clear reduction in human 18S rRNA reads per thousand (left panel, greyscale diamonds) and a 2- to 15-fold increase in parasite reads per thousand (right panel, colored circles). Samples were normalized according to the reads per thousand for reads derived from human host and parasite separately, with the central dotted line reflective of a zero fold change, which marks the undigested samples before treatment with restriction

To quantify the extent to which enzyme digestion improves the limit of detection (LOD) for this method, Plasmodium knowlesi-infected rhesus macaque blood with 3.3% parasitemia (approximately 144,000 parasites per microliter) was utilized. Three aliquots of the sample were serially diluted in parasite-negative whole human blood to a parasitemia of 0.072 parasites per microliter and analyzed in biological triplicate. For the region amplified, the rhesus macaque and human 18S rRNA target sequences are identical, and thus, sample digestion will be equivalent despite the host species being different. Post-digestion samples were, again, cleaned only according to the > 2 kb size selection protocol. As expected, the LOD for samples that had not undergone enzyme digestion before PCR and sequencing was high, such that reads began to fall below baseline at parasitemias between 720 and 72 parasites per microliter; meanwhile, prior enzyme digestion caused samples to fall below baseline between 72 and 7.2 parasites per microliter (Fig. 4a). A trend line was fitted to the log-transformed data to determine a more precise LOD before and after enzyme digestion (Fig. 4b). Using this trend line, the LOD for undigested samples was extrapolated to 163 parasites per microliter while that of the digested samples was estimated at 15 parasites per microliter. As this estimate was extrapolated from a trend line rather than empirical data, a series of finer dilutions (between 61.2 and 0.72 parasites per microliter) was performed to establish a more precise LOD (Fig. 4c). With restriction enzyme digestion, it was confirmed that the assay can detect as few as 7.2 P. knowlesi parasites per microliter of blood, albeit inconsistently, as this LOD was achieved for only two out of three triplicate samples with the third replicate detecting only 28.8 parasites per microliter (Table 2). For the identical set of undigested specimens, a consistent positive result was obtained only at the highest parasite concentration of 61.2 parasites per microliter, with one replicate detecting down to 39.6 parasites per microliter and a second detecting only 50.4 parasites per microliter. Based on this data, there is a 1.4- to 8.5-fold improvement in LOD following restriction enzyme digestion.Fig. 4

Enzyme digestion markedly lowers assay limit of detection. (a) Reads per thousand for undigested (gray) and digested (black) 10-fold serial dilutions of P. knowlesi in whole human blood (n = 4, mean ± SD). (b) Log-transformation of reads per thousand from serially diluted samples suggests a limit of detection of 163 parasites per microliter for undigested samples (gray, r² = 0.9852) and 15 parasites per microliter for digested samples (black, r² = 0.9533) (n = 4 biological replicates, mean ± SD). (c) After deeper analysis, reads per thousand for undigested (gray) and digested (black) serial dilutions between 61 parasites per microliter and 0.72 parasites per microliter demonstrate a limit of detection of 40 to 60 parasites per microliter for undigested samples and 7 to 29 parasites per microliter for digested samples (two-way ANOVA with Sidak’s multiple comparisons posttest, **** p < 0.0001, *** p < 0.001, ** p < 0.005 n = 3, mean ± SD)

Human clinical blood samples used in this study were originally submitted to the CDC Parasitic Diseases Branch for confirmatory diagnosis of parasitic infections. Following diagnosis, samples were de-identified and frozen in 200 μL aliquots at − 80 °C for use in assay development and validation. Samples containing P. falciparum, P. vivax, P. malariae, Plasmodium ovale, B. microti, and Babesia divergens were acquired in this way. For some rare blood-borne parasites, either animal blood or human blood samples collected during previous research studies and stored at − 80 °C were used. Bioinformatic analysis indicated that the restriction enzyme cut sites and the region of amplification were identical to human for all relevant animal samples. Some rare blood parasites that could not be acquired as true clinical samples were recreated by spiking uninfected human blood with cultured parasites—L. donovani subspecies infantum, L. donovani subspecies donovani, T. cruzi, and T. brucei cultures were added to whole human blood at a ratio of 1:10. All blood samples were collected into EDTA anticoagulant except for the clotted blood W. bancrofti sample. The full details of source, matrix, parasite identification, and DNA extraction methods are provided in Table 1.

In an effort to design universal parasite primers, Geneious bioinformatics software (Biomatters Inc., Newark, NJ, USA) was used to create an alignment of the 18S ribosomal RNA genes from the publicly available sequences of 24 protozoa, 17 helminths, and Homo sapiens. No 18S rRNA sequences for M. perstans were available for inclusion in the alignment. Restriction enzyme cut sites were analyzed bioinformatically, and 18 primers were designed and tested in 14 primer combinations with 6 candidate restriction enzymes against a panel of four protozoan and one nematode, one trematode and two cestode helminth parasites (P. falciparum, Toxoplasma gondii, T. brucei, C. felis, Brugia pahangi, Schistosoma mansoni, Dipylidium caninum, and Taenia sp., relatively). From these analyses, one primer set was selected that amplifies an approximately 200 bp region of the 18S rRNA gene in all parasites tested, yielding a clearly visible band on a 1.5% agarose gel. Of the six restriction enzymes tested, BamHI-HF and XmaI (NEB, Ipswich, MA, USA) consistently yielded the highest degree of host DNA removal when paired with the selected primers.

Aliquots of parasite-positive human blood samples at a volume of 200 μL were spiked 1:100 with C. felis-infected cat blood at a concentration of 1.7 million parasites per microliter. DNA was extracted using the QIAamp DNA Blood Mini Kit (Qiagen, Redwood City, CA, USA) and eluted into 50 μL of PCR-grade water. Following extraction, DNA concentrations were determined using a Qubit 2.0 Fluorometer with the Qubit dsDNA High Sensitivity Assay Kit (Life Technologies, Grand Island, NY, USA). One hundred fifty nanograms of DNA were subsequently digested via incubation with 20 units of BamHI-HF and 20 units of XmaI in 1X CutSmart Buffer (NEB) in a final volume of 50 μL for 2 h in a 37 °C water bath. Digested samples were divided into two equal volumes, and enzymes and buffers were subsequently removed using the Monarch PCR & DNA Cleanup Kit (NEB). The Monarch kit enables template size selection (protocol dependent), and since there is uncertainty as to what the template size would be post-digestion, in most cases one sample aliquot was cleaned as a > 2 kb sample, and the second was cleaned as a < 2 kb sample according to the manufacturer’s instructions. Both samples were then eluted in 10 μL elution buffer (NEB).

PCR was performed with the cleaned and digested samples in a reaction volume of 20 μL using Q5 High-Fidelity DNA Polymerase (NEB) according to the manufacturer’s instructions and supplementing with Q5 High GC Enhancer. The primer sequences are as follows: CCGGAGAGGGAGCCTGAGA (forward) and GAGCTGGAATTACCGCGG (reverse). Samples were denatured at 98.0 °C for 2 min followed by 30 cycles of 98.0 °C for 10 s, primer annealing at 67.0 °C for 30 s and extension at 72.0 °C for 45 s, and a final extension at 72.0 °C for 5 min. Following PCR, samples were analyzed on a 1.5% agarose gel, cleaned with the Monarch PCR & DNA Cleanup Kit (NEB, < 2 kb), diluted 1:5 in elution buffer, and transferred to a 96-well plate for library preparation and sequencing.

Library preparation and sequencing were performed by the CDC Biotechnology Core Facility’s Genome Sequencing Lab using the NEBNext Ultra DNA Library Prep Kit for Illumina (NEB), and multiplexing was performed using the NEBnext Multiplex Oligos for Illumina Index kit (NEB) or using TruSeq HT Adapter sets (Illumina). No more than 80 samples were multiplexed on a single MiSeq run to ensure sufficient and consistent sequencing depth for each sample. Runs were prepared using a MiSeq Reagent Nano Kit v2 (PE250bp) (Illumina), and sequencing was performed on the Illumina MiSeq platform (Illumina). For all experiments, the paired digested and undigested (otherwise identical) samples were multiplexed on the same sequencing run to ensure they could be compared directly.

Geneious bioinformatics software (www.geneious.com) was used to analyze the raw fastq files generated by the MiSeq sequencing runs. Reads were paired and subsequently trimmed using the BBDuk plugin with a minimum quality of 35 and a minimum read length of 150 base pairs. Paired and trimmed reads were then mapped to the reference alignment described below using a minimum mapping quality of 35, a minimum overlap of 150, an allowance of zero mismatches per read, and a minimum overlap identity of 99%. To allow comparisons to be made between experiments and to assess the impact on restriction enzyme treatment for host template reduction, samples were normalized by dividing the species-specific mapped reads by the total paired reads in the sample, multiplying that value by 1000, and reporting the final value as reads per thousand. To account for index cross-talk between DNA samples multiplexed on the same Illumina run, a series of careful analyses were conducted to establish both a minimum cutoff to be applied uniformly across all samples as well as a shifting maximum cutoff to be applied on a sample-by-sample basis. Satisfaction of both criteria, as described below, would eliminate the occurrence of false-positive results.

Geneious bioinformatics software was used to establish a set of blood parasite reference sequences from among the 18S rRNA gene sequences found on GenBank. This reference database was used for assay development and validation and utilized the parasite GenBank accession numbers listed in Table 1 and the Homo sapiens sequence available under GenBank accession number HUMRGE. Since the region of amplification shares 100% sequence identity for L. d. donovani/L. d. infantum, T. b. rhodesiense/T. b. gambiense, and W. bancrofti/L. loa, only one representative sequence was included for each of these pairs in the final reference alignment cohort.

A minimum coverage cutoff value of 20 reads was established for the specific protocol described in this study (i.e., using the Illumina MiSeq platform and a 500 cycle Nano Kit, and multiplexing 60 to 80 samples per sequencing run). The following formula was used to determine the minimum coverage cutoff:

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \left[\left({\mu}_{\mathrm{contam}\_\mathrm{all}}\right)+4(S.D.)\right]\times {\upmu}_{\mathrm{sample}\_\mathrm{reads}}={\mathrm{CUTOFF}}_{\mathrm{min}} $$\end{document}μcontam_all+4S.D.×μsample_reads=CUTOFFmin

where μ_{contam_all} = the mean proportion of contaminating reads (i.e., the mean proportion of reads from all blood negative samples, n = 18, from this study that mapped to a parasite reference sequence), S.D. = standard deviation for μ_contam (four standard deviations above the mean was selected as the number of reads obtained for all samples represents a normal distribution), and μ_{sample_reads} = the mean number of reads obtained for each sample from 425 sequenced samples.

For this study, the minimum cutoff was calculated empirically and as follows (using all blood negatives in a sample):\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \left[(0.0001)+4(0.00026)\right]\times 17,553=19.855\ \mathrm{reads} $$\end{document}0.0001+40.00026×17,553=19.855reads

Consequently, specimens were only considered positive if more than 20 Illumina reads were detected for any given parasite species. This represents the maximum number of contaminating reads you might expect for any given specimen regardless of the parasite species under investigation, using this specific protocol.

Similarly, the maximum sliding cutoff was calculated for each individual sample using the following formula:

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \left[\left({\mu}_{\mathrm{run}\_\mathrm{contam}}\right)+4(S.D.)\right]\times {S}_{\mathrm{reads}}={\mathrm{CUTOFF}}_{\mathrm{max}} $$\end{document}μrun_contam+4S.D.×Sreads=CUTOFFmax

where μ_{run_contam} = the mean proportion of contaminating reads within the negative control samples included in this specific sequencing run (at least four negatives were included in each run), S.D. = standard deviation for μ_{run_contam}, and S_reads = the number of reads sequenced for the sample.

These coverage cutoffs take into account the index cross-talk (i.e., sample bleeding) for samples containing parasite reads that were multiplexed on the same sequencing run [25]. Two examples are provided below for calculating CUTOFFmax:

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \left[(0.0002)+4(0.00033)\right]\times \mathrm{10,410}=15.823\ \mathrm{reads} $$\end{document}0.0002+40.00033×10,410=15.823reads

In example A, the sliding maximum rule would suggest a cutoff of 16 reads. However, at this cutoff we cannot be confident that this is not due to index cross-talk [25]. Consequently, we use the value of CUTOFF_min (20 reads) to exclude false positive results. In example B, CUTOFF_max is used (25 reads) to account for the fact that as the number of reads increases (i.e., depth) the proportion of contaminating sequences will also increase. Cutoff values are always rounded up to the nearest whole number.

This dual criterion system was developed to account for certain variables that may affect the sequencing output. As discussed above, at greater sequencing depth, it is more likely that contaminating reads will be detected; the sliding CUTOFF_max accounts for this. Additionally, the composition of samples sequenced in a given run will have an impact on the number and composition of contaminating reads present in negative control specimens as a result of index cross-talk. For example, if a run contains a large number of samples that are positive for P. falciparum, yet a small number of Leishmania positive samples, one would expect to see a larger proportion of P. falciparum reads in the negative control specimens compared to Leishmania reads. This is also the reason why the sliding CUTOFF_max is calculated for each run and for each parasite species individually—it compensates for the diversity of specimens that may be included within and between runs. Furthermore, if a single specimen out of 80 included on a single MiSeq run contains P. knowlesi DNA while all other samples are negative, it is possible that no P. knowlesi reads will be detected in the negative samples. In this case, the sliding CUTOFF_max cannot be used so CUTOFF_min is implemented.

All multiple sequence alignments were performed using the MUltiple Sequence Comparison by Log-Expectation (MUSCLE) algorithm. GenBank accession numbers for human parasites can be found in Table 1, those for vertebrates in Additional file 6: Table S1 and those for fungal organisms in Additional file 8: Table S2.

A dilution series was prepared from DNA extracted from the buffy coat layer of whole blood provided by healthy human volunteers and parasite DNA from 3D7 P. falciparum cultures. Samples were spiked with cat blood infected with C. felis prior to DNA extraction. Dilutions of human DNA were prepared at 3, 2.5, 2, 1.5, 1, 0.5, and 0 ng/μL and supplemented with 0.2 ng/μL DNA from a 3D7 P. falciparum culture (DNA equivalent of ~ 8600 parasite per microliter). Samples were then processed as described above by restriction digestion, PCR enrichment, and deep sequencing. Each digested sample was paired with an identical sample that was not restriction digested. The resulting sequencing reads were mapped against the reference database for quantification of parasite-derived reads. Note that for every experiment, an unquantified proportion of human reads was contributed by human products in the P. falciparum 3D7 culture.

For analysis of assay limit of detection, frozen samples of P. knowlesi from a non-human primate infection for which parasitemia had previously been determined by microscopy at the CDC (~ 144,000 parasites per microliter) were serially diluted in parasite-free whole blood. Samples were processed and sequenced as described above, in triplicate, with restriction digested samples paired with an identical undigested sample sequenced on the same run.

P. knowlesi in rhesus macaque blood was generously provided by Amy Kong (CDC, Malaria Branch, Atlanta, GA, USA); Babesia duncani in gerbil blood and B. divergens stabilate in human blood were kindly provided by Henry Bishop (CDC, Parasitic Diseases Branch, GA, USA); B. malayi microfilariae in feline blood were provided by Andy Moorhead (Filariasis Resource Reagent Resource Center (FR3), Athens, GA, USA); W. bancrofti microfilariae in human blood was provided by Patrick Lammie (CDC, Parasitic Diseases Branch, Atlanta, GA, USA); and C. felis in feline blood was provided by David Peterson (University of Georgia, Athens, GA, USA). L. d. infantum, L. d. donovani, and T. cruzi in culture were generously provided by Marcos deAlmeida (CDC, Parasitic Diseases Branch, Atlanta, GA, USA), T. b. rhodesiense in culture was provided by Stephen Hajduk (University of Georgia, Athens, GA, USA). Finally, L. loa and M. perstans purified DNA was generously provided by Thomas Nutman (National Institutes of Health, Bethesda, MD, USA). Funding for this work was provided by the CDC Advanced Molecular Detection initiative.

Additional file 1:

Figure S1. 18S rRNA Nucleotide alignment showing primers designed to detect a region of the gene wherein XmaI and BamHI restriction enzyme cut sites are present only in in the human host sequence and not in any parasite sequences. (TIF 13161 kb)