An Integrated Multiomics Approach to Identify Candidate Antigens for Serodiagnosis of Human Onchocerciasis*

Improved diagnostic methods are needed to support ongoing efforts to eliminate onchocerciasis (river blindness). This study used an integrated approach to identify adult female Onchocerca volvulus antigens that can be explored for developing serodiagnostic tests. The first step was to develop a detailed multi-omics database of all O. volvulus proteins deduced from the genome, gene transcription data for different stages of the parasite including eight individual female worms (providing gene expression information for 94.8% of all protein coding genes), and the adult female worm proteome (detecting 2126 proteins). Next, female worm proteins were purified with IgG antibodies from onchocerciasis patients and identified using LC-MS with a high-resolution hybrid quadrupole-time-of-flight mass spectrometer. A total of 241 immunoreactive proteins were identified among those bound by IgG from infected individuals but not IgG from uninfected controls. These included most of the major diagnostic antigens described over the past 25 years plus many new candidates. Proteins of interest were prioritized for further study based on a lack of conservation with orthologs in the human host and other helminthes, their expression pattern across the life cycle, and their consistent expression among individual female worms. Based on these criteria, we selected 33 proteins that should be carried forward for testing as serodiagnostic antigens to supplement existing diagnostic tools. These candidates, together with the extensive pan-omics dataset generated in this study are available to the community (http://nematode.net) to facilitate basic and translational research on onchocerciasis.

Onchocerciasis is a neglected tropical disease that is responsible for significant morbidity (blindness and/or severe skin disease) in sub-Saharan Africa. An estimated 37 million people in 34 countries are infected with the causative agent of onchocerciasis, Onchocerca volvulus (1). Large-scale public health programs based on mass treatment with the anthelmintic ivermectin have reduced O. volvulus infection rates in many areas (2,3), and plans are being developed to scale up activities to eliminate the infection (4).
Improved diagnostic tools are needed for onchocerciasis elimination programs to identify hypoendemic areas that were excluded from prior control programs and to determine when transmission has been interrupted. Therefore, the lack of an ideal (i.e. highly sensitive and specific, point-of-care) diagnostic test for adult worms presents a significant hurdle. Identification of worm larvae (microfilariae, MF; Fig. 1) in small skin biopsies ("skin snips") has been the standard diagnostic method for onchocerciasis for many decades. While highly specific, skin snips are insensitive in the case of light infections or infections with worms that are not reproducing (e.g. adolescent worms or worms that have been temporarily sterilized by drug treatments). Several PCR-based DNA detection assays and ELISA-based serodiagnostic assays have been proposed, but these are impractical for use in the field (5)(6)(7)(8)(9). Only one rapid-format serodiagnositic test is available at this time, the S.D. BIOLINE Onchocerciasis cassette test (the "BO TEST") that detects IgG4 antibodies to recombinant antigen Ov16 (10,11).
Serodiagnostic assays like the Ov16 test are particularly useful for monitoring young children toward the end of elimination programs because antibodies should be absent from individuals born after transmission has been interrupted (10). There is a long history of work to develop sensitive and specific serodiagnostic assays for onchocercerciasis (12). In the premolecular era, parasite fractions and extracts were tested for their ability to distinguish the antibody responses of infected and noninfected individuals (13)(14)(15)(16)(17). Later, phage display libraries were screened to identify clones expressing parasite proteins reactive with antibodies in patient sera (18). Several recombinant proteins and protein combinations were assessed (8, 9, 19 -22), and the Ov16 antigen stood out due to its high specificity (6,11,23).
The S.D. Bioline Onchocerciasis antibody test is a promising new tool for community screening and mapping of endemic areas. However, previous studies, including studies of Ov16, have shown that patients with different disease manifestations and in different stages of infection show markedly different antibody responses (10, 24 -27), which could confound even the best single-antigen serodiagnostic assay. Therefore, it would be advantageous to identify additional antigens that could work alone or in combination with Ov16 to further improve diagnostic accuracy. There have been major advances on multiple fronts since Ov16 was discovered in 1991, including methodologies and technologies for antigen identification. Therefore, the goal of this study was to use a multi-omics (genomics, transcriptomics, proteomics, and immunomics) approach to identify novel immunodiagnostic antigens that might lead to improved diagnostic tests for onchocerciasis elimination programs.
RNA Isolation, cDNA Sequencing, and Gene Expression Analyses-Adult O. volvulus worms were isolated from nodules of onchocerciasis patients from the Ashanti region of Ghana in November 2009 and stored at Ϫ80°C until use (40). In total, a pool of 4 male worms obtained from four different patients and 8 individual female worms obtained from 6 different patients were used for transcriptome analysis. RNA was isolated from the pooled males and the individual females using a PureLink RNA Mini Kit according to the manufacturer's suggested protocol (Ambion/Applied Biosystems, Austin, TX). RNA quality and quantity were assessed with an Agilent 2100 Bioanalyzer (Agilent Technologies, Cedar Creek, Texas) and NanoDrop ND-1000 (NanoDrop Technologies, Wilmington, DE), respectively. Total RNA was poly(A) selected using the MicroPoly(A)Purist Kit (Ambion/Applied Biosystems) according to the manufacturer's suggested protocol and reverse transcribed using the Ovation RNA-Seq V2 kit (NuGen Technologies, Inc., San Carlos, CA) with random and poly(A) primers. Paired-end cDNA libraries were generated and sequenced on the Illumina HiSeq 2000 platform according to standard protocols, and raw reads were submitted to the GenBank sequence read archive (SRA) under BioProject PRJNA219638 (supplemental Table S2). Sequence data available from the SRA for O. volvulus MF, L3 and adult male (accession numbers ERX200391-ERX200394, ERX200396, ERX200397) were downloaded and also employed in this study.
Relevant adapter sequences and low quality regions were trimmed, and reads were filtered based on length, complexity, and similarity to suspected contaminants as previously described (41). Remaining, high-quality reads were aligned to the O. volvulus genome assembly using Tophat2 (version 2.0.8, default parameters, (42)) using the genome annotation (gff3 file) as a guide. The number of reads associated with each gene was tallied using HTSeq-Count. Normalized transcript expression levels were calculated using gene lengths and read counts from HTSeq-Count output (fragments per kilobase per million reads mapped; FPKM). Genewise expression levels can be found in Table S1 for all genes detected by LC-MS.
Preparation of Soluble Parasite Proteome-Soluble O. volvulus protein extract was prepared form a pool of 3 adult female worms isolated from one patient in Bong County, Liberia in 1988 (courtesy of the late Dietrich W. Bü ttner) as previously described (41). Briefly, the worms were processed in a 1 ml Dounce homogenizer (GPE Scientific Limited, Leighton Buzzard, UK) in RIPA buffer (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 0.2% sodium deoxycholate, 1 mM EDTA and 10 mM NaF). The homogenate was centrifuged, and protein in the supernatant was measured using the Pierce BCA assay (Thermo Fisher Scientific, Rockford, IL).
Molecular Weight Fractionation of Adult Female Protein Lysate-Adult female lysate (450 g) was separated into eight molecular weight fractions (5-150 KDa) using a GELFrEE 8100 fractionation system with an 8% cartridge (Expedeon, San Diego, CA) (43,44). Protein fractions were precipitated using an acetone-based method and re-solubilized in 100 mM Tris-HCL pH 8.5. The protein and peptide quantities were determined using the Advanced Protein Assay kit (Cytoskeleton, Inc., Denver, CO) and a Qubit® 3.0 Fluorometer. The fractionation was also assessed using SDS-PAGE in MES running buffer (4 -12% Criterion XT gels, BioRad, Hercules, CA). The highest molecular weight fraction (Ͼ120-KDa) did not contain detectable protein and was not analyzed further. The proteins in GELFrEE fractions (ϳ20 -120 g) were precipitated twice using 3ϫ volumes of cold acetone.
Preparation of Human IgG and Immunoaffinity Purification of Adult Female Worm Proteins-A pool of 10 sera (300 l each) from O. volvulus Mf and nodule positive, de-identified patients were used in immunoaffinity purification assays. Eight sera were collected in Feb-ruary 2014 from patients living in Lofa County Liberia, and two sera were collected from Loa loa and Mansonella perstans Mf negative patients from Kumba, Cameroon in November 1997. All onchocerciasis patient sera tested negative for circulating W. bancrofti antigen by the Binax ICT card test. A pool of 3 sera (1 ml each) from healthy individuals from Missouri collected in 2010 was used as negative control.
Immunoaffinity purifications were carried out as previously described (41). Briefly, total IgG was precipitated from pooled onchocerciasis patient sera and pooled control sera, respectively, and coupled to Pierce NHS-active agarose (Thermo Fisher Scientific) in spin columns (Thermo Fischer Scientific). Columns were washed, blocked, and incubated with 600 mg O. volvulus soluble protein overnight at 4°C. Columns were thoroughly washed, and immune complexes were eluted with Pierce IgG elution buffer (Thermo Fisher Scientific) in 1 ml fractions. Fractions were neutralized with 1.0 M Tris pH 9.0, and aliquots were analyzed by Western blot using the original pooled patient or control sera as the primary antibody. The fraction (ϳ1.5 ml) with the greatest Western blot reactivity from each purification column was concentrated to ϳ100 l in a MICROCON® (YM-3) device (Millipore, Darmstadt, Germany) for protein digest as described below.
Protein Digestion and Peptide Purification-Protein pellets from acetone precipitation of molecular weight fractions of female worm lysate were dissolved in 20 l of Tris buffer (100 mM, pH 8.5) containing 8 M urea. As a digestion standard, horseradish peroxidase (1 g) was added to each sample. The proteins were reduced using TCEP (5 mM, Thermo Fisher Scientific) for 30 min, and alkylated with iodoacetamide (40 mM) (Sigma-Aldrich, St. Louis, MO) at room temperature in the dark for 30 min. The reactions were quenched with DTT (20 mM) (Sigma-Aldrich) for 15 min. The samples were digested with endoprotease Lys C (5 g) (Sigma-Aldrich) for 30 min at 37°C in a Barocycler (Pressure BioSciences, South Easton, MA), followed by a fourfold dilution with buffer, addition of trypsin (5 g) and incubation at high pressure for 30 min at 37°C. The digests were acidified with 1% TFA and the peptides were purified using SepPak cartridges and elution conditions as previously described (45). The eluted peptides were dried in a SpeedVac and dissolved in water/acetonitrile/formic acid (98%/1%/1%) and transferred to autosampler vials (SUN-SRI, Rockwood, TN) for storage at Ϫ80°C prior to LC-MS analysis.
The immunoaffinity enriched proteins (ϳ 1 ml) were concentrated in a MICROCON® (YM-3) device (Millipore) to ϳ 100 l for filter-aided sample preparation (46). The concentrate was transferred to a MICROCON® (YM-30) device with multiple washes with Tris buffer (100 mM, pH 8.5) containing 8 M urea. After two Tris buffer exchanges (200 l), the protein was reduced with 100 mM DTT at room temperature for 30 min and reduced with iodoacetamide (50 mM) for 30 min. The reduced and alkylated proteins were exchanged into a volatile buffer for digestion (ammonium bicarbonate, pH 7.4). Trypsin (1 g; Cleavage after Lysine (K) or Arginine (R) except when either is followed by proline (P)) was added and digestion at 37°C was performed overnight in a Thermomixer (Thermo Fisher Scientific). The digest was acidified to 5% formic acid. The digests were desalted using NuTips (Glygen, Columbia, MD) with sequential extraction with C 4 and porous graphite carbon tips on a Biomek NXP robot (Beckman Coulter, Pasadena, CA). The eluates (70% acetonitrile) from the two tips were combined, concentrated to near dryness and dissolved in acetonitrile/formic acid (1%/1%) for LC-MS analysis or storage at Ϫ80°C.
Data acquisition was performed with a TripleTOF 5600ϩ mass spectrometer (AB SCIEX, Concord, ON) fitted with a PicoView Nanospray source (PV400, New Objectives, Woburn, MA) and a 10 m Silica PicoTip emitter (New Objectives) for bottom-up proteomics. Data were acquired using an ion spray voltage of 2.9 kV, curtain gas of 20 PSI, nebulizer gas of 25 psi, and an interface heater temperature of 175°C. The MS was operated with a resolution of greater than or equal to 25,000 (fwhm) for TOF-MS scans. For data dependent acquisition, survey scans were acquired in 250 ms from which 100 product ion scans were selected for MS2 acquisition for a dwell time of 100 ms. Precursor charge state selection was set at ϩ2 to ϩ5. The survey scan threshold was set to 100 counts per second. The total cycle time was fixed at 2.25 s. Four time bins were summed for each scan at a pulser frequency value of 15.4 kHz through monitoring of the 40 GHz multichannel TDC detector with four-anode/channel detection. A rolling collision energy was applied to all precursor ions for collision-induced dissociation as described in the Analyst software.
The (2) peptide fragment ion tolerance of 100 ppm (the larger error allowed for MS2 fragment identifications was used to capture lower intensity fragment ions that may have fewer detector events for accurate determination of the center of the mass measurement); (3) trypsin cleavage specificity with up to three missed cleavages and a single semitryptic peptide per sequence entry; (4) variable oxidized modification of Met and constant modification of Cys (carbamidomethylaton). Quality peptide spectral matches with the MS2 high resolution scans were determined with a false discovery rate threshold of 1% using a decoy fusion database algorithm (47). Identifications were made with Ͻ0.1% False Discovery Rate (FDR) at the protein level. Individual spectra required a minimum PEAKS score of 20 to be accepted, according to the software manual recommendations. Protein identification required at least two unique peptides sequences, not considering modifications or isobaric sequences. The inferred proteins, their quantification, and their protein group accessions for the worm proteomic study with molecular weight fractionation and the immunoaffinity enrichment study are provided in supplemental Data sets S1 and S2, respectively. The index scan numbers for viewing the MS2 spectra is provided in these tables, along with accession numbers for the proteins. Single peptide and PMF data are deposited in PeptideAtlas (PASS00679), and the complete protein list (along with the number of peptides assigned to each detected protein) are in supplemental Table S1.
Experimental Design and Statistical Rationale-All life cycle stages analyzed by RNAseq were represented by at least two biological replicates (8 individual adult females, 3 for L3, 2 for adult males, and 2 for MF). This allowed for the confirmation of consistency of expression and to measure variability in expression. Statistical analysis of differential expression at the RNA level was calculated using standard settings in DESeq2 (48) considering all available biological replicates.
Whole-worm lysate was separated into 8 molecular weight fractions (5-150 kDa) for proteomics analysis in order to better capture the overall proteome. The immunoaffinity purified proteomics samples included an O. volvulus-infected sample as well as a noninfected sample to serve as a control for background protein detection, and the whole-worm lysate additionally served as a positive control to identify likely worm-derived proteins. The samples analyzed in our proteomic studies were single-replicate; however, this data was not used to infer statistical differences in protein abundance, rather to identify presence or absence of the proteins in the different samples.
Significant enrichment and depletion of deduced proteins with various properties among protein sets was tested using a cumulative nonparametric binomial distribution test (MS Excel version 2011); property occurrence rates in the whole genome were used as the background set in all cases. FDR correction of p values (minimum threshold 0.05) was used to correct for multiple testing, in cases where multiple tests were ran for a single enrichment test (49).
Ethics Statement-All worm specimens were untreated worms collected during a chemotherapy trial for which proper IRB approval was available. We have no information linking the worms to individual patients. The use of de-identified patient sera for the development of new diagnostics was approved by the Washington University School of Medicine IRB under the protocol number 201102546.

RESULTS AND DISCUSSION
Parasite Material-This study placed a particular emphasis on adult female worms for several technical and biological reasons. Technically, the adult females are large (ϳ40 cm) enough to yield sufficient material for RNAseq and proteomic analyses, and relatively easy to isolate compared with other stages because they reside in subcutaneous nodules (Fig. 1). Second, they are responsible for a significant fraction of worm excretory/secretory products, presumably due to the process of birthing offspring (50), which may trigger an antibody response from the host. Third, they contain developing offspring, so microfilarial antigens will be represented in adult female RNA and lysate to some extent.

Annotation and Conservation of O. volvulus Predicted
Proteins-The identification of proteins by mass spectrometry relies on a sequence database searching approach wherein acquired MS2 spectra are matched to database peptide sequences after in silico endoprotease digestion (Fig. 2). The predicted protein sequences from an unpublished draft version of the O. volvulus genome (WormBase WS245) were used for our analyses. Functional annotations were inferred based on sequence similarity (supplemental Table S1).
Primary sequence similarity searches were used to determine the level of conservation shared between proteins from O. volvulus and relevant species (i.e. filarial nematodes and soil-transmitted nematodes with overlapping geographical distributions, the human host, and representative bacterial and yeast species; see Methods). This more targeted approach was preferred over searches against NR because NR contains closely related species that are not pertinent to our search for an appropriate serodiagnostic antigen (e.g. Onchocerca parasites of cattle). In total, 4753 inferred proteins were considered O. volvulus-specific in this analysis using a cutoff of 70% sequence identity over 70% sequence length (supplemental Fig. S1, supplemental Table S1).

RNA-level Expression of Genes Encoding O. volvulus Proteins-An
RNAseq approach was used to assess the expression levels of predicted O. volvulus genes among adult worms (8 females and a pool of males). Comparison of global gene expression among these samples indicated a vast difference between males and females, but also a degree of diversity among the females (supplemental Fig. S2). The expanded RNAseq analysis of the eight adult female worms indicated expression signals for 11,508 of 12,143 genes (94.8%). Of those, 7361 genes had expression signals in all eight of the individual females studied (with a minimum breadth of RNAseq read coverage Ն50%); 2820 of these genes were considered to have consistently high expression (with the minimum expression level among the eight adult females being in the top 25% of all genes) and low variance (with less . Over the course of a few months to a year, the worms molt (shed the outer cuticle) twice and develop into sexually mature adults. In cooperation with the host immune system, the female worms form fibrous nodules in the skin (indicated with red dots). The females remain sessile inside these nodules indefinitely whereas adult males migrate between the nodules to mate. Patent females can release thousands of microfilarial offspring per day, which migrate through the skin until they are picked up by a biting blackfly. The larvae molt twice inside the fly before reaching the infective stage. Light-orange boxes indicate stages for which RNAseq data was available and used in the current study. than 25% standard error, relative to the average expression value; supplemental Fig. S3; supplemental Table S1).
Proteomic Analysis of Adult Female Lysate-Soluble protein from mature, adult female worms was fractionated and subjected to analysis by liquid chromatography-MS (LC-MS) 1 in order to assess expression at the protein level. We identified 24,898 unique peptide sequences that were mapped onto 2126 O. volvulus proteins (supplemental Table S1), a number consistent with our expectations based on previous analyses of total parasitic worm lysates using similar methods (41,51,52). These proteins were enriched for signal peptides for secretion and depleted for transmembrane domains (p Ͻ 10 Ϫ5 and p ϭ 2 ϫ 10 Ϫ5 , respectively). Spectral counts were used to obtain an estimate of protein abundance. As expected, there is a degree of correlation between transcript and protein expression levels (Pearson correlation ϭ 0.20, p Յ P Ϫ 5 ; compared with 0.20 in the parasitic roundworm A. suum (53) and 0.19 in yeast (54)), and proteins detected in the worm lysate were likely to be represented at relatively high expression levels in the RNAseq datasets (Figs. 3A  and 3B).
Identification of Immunoreactive Proteins-Immunoreactive O. volvulus proteins were purified from adult female lysate by immunoaffinity enrichment with IgG from human sera and identified after endopeptidase digestion and LC-MS (supplemental Table S1). A total of 248 proteins were found among those bound by pooled onchocerciasis patient IgG but not IgG from healthy North Americans (Fig. 2). All but seven of these were also found in the whole worm proteome (supplemental Table S1), and they tended to be detected at relatively high expression levels in the female worm RNAseq data and in the whole worm proteome (Figs. 3B and 3C). This approach identified many of the serodiagnostic targets that have been proposed and characterized over the past 25 years (Table I).
Oddly, Ov16 and Ov33, two of the major serodiagnostic antigens for onchocerciasis, were not included in our list of high priority antigens because they were bound by control human IgG. This cross-reactivity could be due to the use of total IgG in immunoaffinity purifications rather than IgG4. Many available serodiagnostic assays, like the Ov16 rapid test and Ov33 ELISA (9, 11) measure IgG4 antibodies because this antibody subclass provides higher specificity than total IgG for helminth infections (55). Otherwise, there must be a quantitative difference in the titer of reactive IgG in infected patients compared with uninfected controls because both of these antigens have proven useful in diagnostic assays.
Characterization of Putative Diagnostic Antigens-Proteins of interest from the immunoprecipitation study (n ϭ 241; Fig.  2) were further characterized based on specific properties desirable for serodiagnostic antigens as outlined in Fig. 2. Although a diagnostic test should be both sensitive and specific, specificity is particularly important for tests used in community-wide screening. Therefore, it is advantageous to select O. volvulus proteins that are not highly conserved with orthologs in relevant species (i.e. humans and other parasites that commonly infection humans in O. volvulus endemic areas). One hundred eighty one of the 241 immunoreactive proteins shared greater than 70% amino acid sequence identity over more than 70% of the total protein length with orthologs from relevant species. It is difficult to predict antibody cross-reactivity based on global sequence similarity, but this high level of conservation makes these proteins less attractive as candidate immunodiagnostic antigens than those that are 1 The abbreviations used are: LC-MS, high performance nanoliquid chromatography interfaced to a high-resolution hybrid quadrupole-time-of-flight mass spectrometer; RNAseq, high throughput sequencing of mRNA; KEGG, Kyoto Encyclopedia of Genes and Genomes; NR, GenBank nonredundant protein database; FDR, false discovery rate. less conserved. Of the 60 remaining protein candidates (supplemental Table S3), 51 were expressed at detectable levels in all eight of the female worms in the RNAseq arm of this study, and 33 of these showed consistently high expression among the individual worms (Fig. 4, supplementary Table S1).
Discussion of Select Candidate Antigens in Light of Presented Multi-omics Data-The 33 O. volvulus-specific and consistently expressed serodiagnostic antigens highlighted in this study are listed in Fig. 4. While several diagnostic antigens previously described in the literature were identified in the immunoprecipitation study, most of these failed in our prioritization scheme due to high levels of sequence conservation with orthologs in relevant species (i.e. Ն70% ID over Ն70% length), particularly W. bancrofti and L. loa, two filarial nematodes with geographical distributions that overlap that of O. volvulus (Table I). As previously mentioned, Ov16 and Ov33 failed our prioritization because they were present among the antigens pulled-down by control human IgG. Though Ov33 did not meet our blast specificity cutoffs, Ov16 would have qualified for our priority list if it had not been bound by control IgG. Only two previously described antigens (RAL-2 and Ov7) passed through all of our prioritization filters.
RAL-2 was originally identified as a putative vaccine candidate because it was recognized by rabbit antibodies raised against O. volvulus infective larvae (18). A later study showed that 88% of people with O. volvulus microfiladermia had antibodies reactive with RAL-2, but no specificity results were reported (24). Ov-RAL-2 shares 58% sequence identity at the amino acid level with its closest ortholog, the Bm14 antigen from B. malayi that has been widely used for serodiagnosis of lymphatic filariasis (56,57). Because sera from some onchocerciasis patients contain antibodies reactive with Bm14, it is possible that O. volvulus RAL-2 will also have specificity issues (56). Further testing will be needed to determine if this is the case.
Like RAL-2, Ov7/OC9.3 was identified by screening a cDNA expression library with sera from infected patients and animals (20,58). This antigen showed promising specificity, but sensitivity was in the range of 81-84% in patients with microfilaremia (20,59,60). Although RNAseq data indicate peak expression of this gene in the L3 stage, one study of experimentally infected chimpanzees showed that antibodies against OC9.3 were usually only detectable after the onset of microfilaremia (60); however, the kinetics of the antibody response may be different in humans.
Although the remaining 31 new candidates met our cutoffs for global sequence similarity to orthologs from relevant species, several were found to contain smaller regions of high sequence conservation upon closer inspection. For instance, portions of OVOC9752 (the candidate with second-greatest abundance in the total proteome) share very high sequence identity (95% ID over 42.6% length) with a thioredoxin peroxidase from W. bancrofti. Similarly, OVOC8985, contains an N-terminal region with high homology (95% ID over the first 230aa) to triosephosphate isomerases from several filarial species. These proteins may still make attractive serodiagnostic antigen candidates, as it is possible to express recombinant peptides that represent only the unique region(s) of the parent sequence. However, this should be considered with caution, as the conserved region could contain the epitope(s) that interact with patient IgG.  4. Characterization of 33 highlighted serodiagnostic candidates. KEGG annotations, stage-specific transcript expression levels, abundance in total worm proteome, and phylogenetic conservation of all 33 prioritized serodiagnostic candidate proteins. In applicable cases, global percent ID and percent length were summed over multiple high-scoring segment pairs. Among our novel candidates, OVOC4612 showed relatively high expression in the total worm proteome and relatively low sequence conservation with relevant orthologs (supplemental Fig. S4). Like the Ov16 antigen, RNAseq data indicates that peak transcription of this gene occurs in adult worms (Fig. 4). Though studies of infected chimpanzees indicated that antibodies against Ov16 were detectable prior to the appearance of MF in the skin, they still took several months to appear (61). Similar RNAseq expression profiles do not guarantee similar protein expression profiles or immune responses, but it is possible that antibodies against this protein could also take several months to become detectable considering the timing Ov16 seropositivity. In contrast, antibodies against OC3.6/ Ov33 are detectable much sooner after infection compared with antibodies against Ov16 (61). RNAseq data indicates that OC3.6 shows peak expression in adult females, but it is also highly expressed in L3. OVOC2486, another novel candidate with low sequence conservation and a putative ShTK domain (supplemental Fig. S5), has an expression profile similar to OC3.6 ( Fig. 4), so it is possible that antibodies against this protein could appear sooner than those targeting proteins that are not strongly expressed until later in the life cycle.

CONCLUSIONS
This project demonstrated how a multi-omics approach can be used to efficiently identify parasite candidate antigens that may be useful for serodiagnosis. The knowledge-based prioritization scheme that we employed (limited similarity to orthologs in other parasites or in humans, expression in relevant parasite life cycle stages, and consistent expression in individual adult worms) illustrates one way to select candidates that warrant further investigation. The comprehensive database developed during this study will allow us and others in the scientific community to explore different prioritization criteria to select potential targets for diagnosis, drugs, or vaccines as they see fit. We plan to validate our prioritization schema by studying the candidate antigens' reactivity with antibodies in individual serum samples from patients with onchocerciasis and other nematode infections. Thus, future research will determine the utility of proteins identified in this study (alone or in combination with other proteins) for diagnosis of individual patients or for mapping or monitoring onchocerciasis elimination programs.