Abstract
Background
Shigella causes severe diarrheal disease, and S. flexneri and S. sonnei are the targets for multivalent vaccine development. Culture-based agglutination has been the gold standard for serotyping, but it is limited by logistics, subjectivity, and the availability of antisera for emerging serotypes. Newer methods, including a real-time PCR-based approach and whole-genome sequencing offer alternatives, but their performance in Shigella endemic populations are not well documented.
Methods
Shigella isolates obtained from the Enterics for Global Health (EFGH) study in Iquitos, Peru were simultaneously serotyped using four methods: culture-based agglutination, isolate-based real-time PCR serotyping, stool-based real-time PCR serotyping and WGS using the in-silico tool ShigaPass. The definitive adjudicated serotype was established by an expert analysis of the WGS data, involving the mapping of sequence reads to known O-antigen biosynthesis and modification genes to identify key mutations.
Results
Results from all four serotyping methods were available for 107/114 isolates. Accuracy for vaccine subtypes S. flexneri 1b, 2a, 3a, 6, and S. sonnei, ranged from 93.3-100% for all methods. Complete concordance between methods was noted in 83/107 isolates, while 24/107 (22.4%) exhibited at least one discrepancy. Most discrepancies derived from S. flexneri serotypes Y, Yv and 1a. Agglutination misclassified eight Y/Yv isolates as 4a, and six isolates correctly classified as 1a by agglutination were classified as 1b by the other methods, a discrepancy associated with a nonsense mutation in the oac gene.
Conclusion
All four serotyping methods achieved acceptable accuracy for Shigella vaccine efficacy evaluation. Although discrepancies are infrequent, WGS provides information of their genomic basis.
Author summary
Shigella serotyping is critically important for the evaluation of future multivalent vaccines, of which there are several in advanced stages of development, as well as for monitoring of emerging Shigella serotypes. Culture based agglutination is the most widely used serotyping method, yet its successful implementation is associated with key logistical constraints. This study compares culture-based, real-time PCR-based, and whole-genome sequencing serotyping methods using isolates from a Shigella-endemic population in Peru. The study demonstrates that molecular and genomic approaches achieve high accuracy for vaccine-relevant serotypes and identifies the genomic basis of serotyping discrepancies. These methods would also reduce variation and improve data quality for future vaccine trials and epidemiologic surveillance. Ultimately, this work informs clinical microbiology laboratories and public health programs that seek a reliable and scalable alternative to traditional serotyping methods.
Introduction
Diarrheal diseases continue to contribute to 500,000 deaths annually, globally [1]. Shigella spp. is estimated to be the leading cause of death due to diarrhea as a result of a bacterial pathogen, and several vaccines are currently being developed for its control and prevention [2–4]. The majority of cases of Shigella are caused by Shigella flexneri (comprised of 16 serotypes) and S. sonnei (comprised of a single serotype), and current strategies aim to develop multivalent vaccines with antigens restricted to these two serogroups [5].
Vaccine development and evaluation require robust and durable typing strategies for organisms such as Shigella flexneri that have multiple serotypes. The lipopolysaccharide (LPS) of Shigella consists of a relatively well conserved core lipid A structure, a species specific O-antigen backbone of polysaccharides encoded by the rfb genes, and a diverse set of terminal O-polysaccharide modifications that originate from lysogenic phages and plasmids [6–13]. The terminal O-antigen modifications account for the antigenic specificity described by current typing strategies [14].
The culture-based isolation and typing of Shigella using commercially available reagents has been the historic gold standard. New typing strategies include a real-time PCR-based approach directly either on cultured, single Shigella isolates or on stool samples, the latter of which provides advantages such as batched and centralized processing. This approach targets the gtr and oac O-antigen modification genes for serotyping S. flexneri [15–17], and holds several advantages over traditional microbiology including increased sensitivity of Shigella detection, batched and centralized sample processing, and reduced variability of culture due to differential abilities and laboratory capacities.
This study utilized data from the Enterics for Global Health (EFGH) site of Peru that concurrently serotyped Shigella using agglutination, a real-time PCR typing strategy on both cultured, single isolates and stool samples, and whole genome sequencing (WGS). Sequence data was analyzed using publicly available in silico tools and expert analysis to compare the performance of each method. Typing schemes were compared based on their ability to correctly characterize Shigella isolate serotypes.
Materials and methods
Ethics statement
The study was approved by Ethics Review Committee of Asociacion Benefica Prisma and the Institutional Review Board of the University of Virginia. Written informed consent to participate in the study was obtained from the parents or legal guardians of children. Participants consented for further use of biological specimens.
Sample origin
Shigella isolates were obtained as part of the EFGH study of Peru. This is an observational cross-sectional study designed to determine the incidence of medically attended shigellosis in children six to 35 months of age from seven different sites located in resource limited communities [18,19]. The EFGH study of Peru was conducted in Iquitos, Peru, between 2022 and 2024. Details of the site and population have been described previously [20].
Serotyping methods
Culture and Agglutination based Serotyping:
The microbiology methods of the EFGH study have been detailed previously [21]. Briefly, rectal swabs were obtained using two nylon flocked swabs and each individually placed in Cary Blair and modified BGS transport medium (mBGS). Samples were transported to the laboratory for processing within 16 hours of collection at 2-8oC. Rectal swabs were used to inoculate both MacConkey (MAC) and XLD agar plates and incubated at 37oC for 24 hours. Non-lactose fermenting colonies with morphological characteristics compatible with Shigella were picked from both plates and re-streaked in clean fresh plates before initiating the biochemical test battery. Colonies were inoculated into motility indole ornithine (MIO), Kligler’s iron sugar (KIA), lysine iron agar (LIA) and urea agar. Colonies with Shigella compatible biochemistry were serotyped by agglutination with Denka Seiken polyvalent and S. flexneri monovalent antisera. Measures to ensure the quality control and reproducibility of agglutination-based serotyping have also been described [21].
ii. Isolate based real-time PCR-based serotyping:
DNA extraction was completed from Shigella isolates using the crude lysate method. Specifically, 10 Shigella colonies cultured in trypticase soy agar (TSA) of each isolate were suspended in 200uL of 1mM EDTA buffer. Samples were incubated at 95oC for 15 minutes, centrifuged at 5000X g for 10 minutes and the supernatant was transferred into a new 1.5mL microcentrifuge tube for storage until real-time PCR testing was performed for Shigella serotyping and speciation as previously described [22]. Briefly, multiplex real-time PCR testing was performed using the Ag-Path-ID One Step RT-PCR Kit (Thermo Fisher, Waltham, MA) in an QuantStudio 7 real-time PCR instrument (Applied Biosystems, Waltham, MA) under the following cycling conditions: 45 °C for 20 minutes, 95 °C for 10 minutes and 40 cycles of 95 °C for 15 seconds and 60 °C for 1 minute. Gene targets are shown in Table 1, and primers and probes in S1 Table [22].
Table 1. Real-time PCR panels used in the current study and the corresponding Shigella species and S. flexneri serotypes. The colours indicate the fluorophore of the probes, blue for FAM, green for VIC, red for Texas Red, and purple for Cy5.
| Shigella species or S. flexneri serotype | Real-time PCR panel I | Real-time PCR panel II | Real-time PCR panel III | |
|---|---|---|---|---|
| S. flexneri 1a | gtrI | |||
| S. flexneri 1b | oac | gtrI | ||
| S. flexneri 1d | gtrX | gtrI | ||
| S. flexneri 2a | gtrII | |||
| S. flexneri 2b | gtrII | gtrX | ||
| S. flexneri 3a | oac | gtrX | ||
| S. flexneri 3b | oac | |||
| S. flexneri 4a | gtrIV | |||
| S. flexneri 4b | oac | gtrIV | ||
| S. flexneri 5a | oac | gtrV | ||
| S. flexneri 5b | oac | gtrX | gtrV | |
| S. flexneri 6 | wzx6 | |||
| S. flexneri 7a | gtrI | gtrIc | ||
| S. flexneri X | gtrX | |||
| S. sonnei | S. sonnei | |||
| Shigella | ipaH | ipaH3 | ||
Serotype assignment was performed based on the following algorithm based on the optimization of the original scheme through EFGH [23]:
The Cq of the S. flexneri serotyping target or S. sonnei target must be within 7.5 Cq of the ipaH Cq.
If ≥2 targets are required to determine the serotype, the Cq difference between the targets must be ≤ 6 Cq.
As noted in Table 1, this algorithm can identify Shigella spp., S. sonnei, S. flexneri, and S. flexneri serotypes: 1a, 1b, 1d, 2a, 2b, 3a, 3b, 4a, 4b, 5a, 5b, 6, 7a, and X. It cannot, however, identify S. flexneri Y, which is characterized by the absence of O-antigen modifications.
iii. Stool based real-time PCR-based serotyping:
A third FLOQSwab rectal swab was obtained and placed in a dry tube for molecular identification of Shigella spp. These methods have also been previously detailed [23]. Briefly, total nucleic acid (TNA) was extracted from rectal swabs using the QIAamp Fast DNA Stool mini kit (Qiagen, Hilden, Germany) with a pretreatment that included bead beating and 95°C incubation. TNA is then eluted with 200 μL of elution buffer. External controls, including 106 phocine herpes virus (PhHV) and 107 MS2 bacteriophage were spiked into each sample during the initial lysis step to monitor the extraction and amplification efficiency. One extraction blank was included per batch of extraction to assess for contamination. Samples were run using a TaqMan Array card with 82 targets, including Shigella species, and S. flexneri serotyping targets in a QuantStudio 7 real-time PCR instrument (Applied Biosystems, Waltham, MA), which included the same gene targets, primers and probes as described in the isolate real-time PCR based serotyping (S1 Table). The assay master mix was prepared using Ag-Path-ID 2X RT-PCR buffer and Ag-Path-ID enzyme mix (Thermo Fisher, Waltham, MA), in addition to the TNA template. Cycling conditions were as follows: 45°C for 20 minutes and 95°C for 10 minutes, followed by 40 cycles of 95°C for 15 seconds and 60°C for 1 minute. Serotype assignment for attributable Shigella detection (ipaH Cq < 29.5) was performed as described for the isolate based real-time PCR serotyping [19,22]. Additionally, if multiple S. flexneri serotypes and/or S. sonnei are detected using the above criteria, the target(s) with the lower Cq determines the primary species present.
Fig 1 shows the S. flexneri serotype O-antigens, its modifications, and the genes utilized in real-time PCR for serotyping.
Fig 1. Shigella flexneri O antigen modifications and associated genes.
Agglutination, stool-based real-time PCR and isolate-based real-time PCR were run in duplicate for all samples for which a discrepant result was obtained.
iv. Sequencing and in-silico genome-based prediction of serogroup and serotype:
Genomic DNA from all isolates was purified using the Wizard genomic DNA purification kit according to the manufacturer’s directions (Promega, US). Genomic libraries were prepared using the Illumina DNA Prep Tagmentation kit as described previously [24] and sequenced using a MiSeq Reagent Kit v2 (500-cycles) on a MiSeq instrument (Illumina) at 16 pM, following the manufacturer’s protocols. Draft genomes are available at NCBI and are associated with BioProject PRJNA1321318 and accession numbers are in S2 Table.
Reads were uploaded to Enterobase (http://enterobase.warwick.ac.uk/) [25] for assembly, quality control statistics and basic analysis. Additionally, assembled contigs were run through CheckM software v1.1.3 for completeness and contamination [26]. Assembled contigs were run through ShigaPass v1.5.0, an in-silico tool to predict Shigella serotypes from whole genome assemblies [27].
v. Manual genome assessment serotype adjudication:
Manual genome assessment for serotype adjudication was based on presence/absence of S. flexneri O-antigen modification genes and assessment of gene integrity. Detection of O-antigen modification genes and identification of premature stop codon mutations indicative of gene inactivation were performed using the Map to Reference function in Geneious Prime (v2025.1.2; Biomatters, Ltd., Auckland, New Zealand). Illumina paired-end reads >150 nt generated from an individual S. flexneri strain were mapped to 8 O-antigen biosynthesis and modification regions: wba (AE005674 (SF2096-SF2104)), gtrABI (CP130063 (QS169_18520-QS169_18530)), oac1b (JN377795), gtrABII (AE005674 (SF0305- SF0307)), gtrX (L05001), oac (AF547987), gtrIV (NC_022749 (V416_gp25)), and opt (KC020049). All S. flexneri serotypes 1–5, X and Y should have sequence reads that map across the wba locus, which encodes S. flexneri O-antigen backbone biosynthesis enzymes. Both S. flexneri 1a and S. flexneri 1b strains should have sequence reads that map to the gtrABI locus, and S. flexneri 1b should have sequence reads that map to oac1b. Both S. flexneri 2a and S. flexneri 2b strains should have sequence reads that map to the gtrABII and S. flexneri 2b should have sequence reads that map to gtrX. Both S. flexneri 3a and S. flexneri 3b strains should have sequence reads that map to oac, and S. flexneri 3a strains should have sequence reads that map to gtrX. S. flexneri 4a, S. flexneri 4av and S. flexneri 4b strains should have sequence reads that map to gtrIV, S. flexneri 4av strains should have sequence reads that map to opt, and S. flexneri 4b strains should have sequence reads that map to oac. S. flexneri Y strains have reads that only map to the wba locus and S. flexneri Yv strains should have sequence reads that map to opt. Genomic assemblies with ≥5 × coverage across O-antigen modification genes were considered sufficient for two assessments: detection of gene presence or absence, and identification of premature stop codons arising from frameshift or nonsense mutations. Stop codon calls were accepted only when the causative mutation was present in all mapped reads at that position, confirming a true chromosomal mutation rather than a sequencing artefact.
Comparative analysis
The identification of S. flexneri O-antigen modification genes and mutations performed in Geneious Prime (v2025.1.2; Biomatters, Ltd., Auckland, New Zealand) was considered the method to overrule discrepancies and definitive technique to adjudicate Shigella serogroups and serotypes. Agglutination, real-time PCR (on isolates and stool samples) and ShigaPass were compared to each other and ultimately to the gold standard.
A “negative” result is defined as the absence of the detection of Shigella in the specific sample analyzed. A “not typed” result is defined as a positive Shigella sample for which the test could not assign a serotype. The accuracy of each typing method was calculated as the total number of matching serotypes over the total number of samples evaluated by both compared methods, with 95% confidence intervals (CI) estimated using the Wilson score method. Additionally, the accuracy of each of the four serotyping methods for the Shigella serogroups and serotypes contained in the three main candidate vaccines (candidate A: S. sonnei, S. flexneri 1b, 2a and 3a (2); candidate B: S. sonnei and S. flexneri 2a. (3); candidate C: S. sonnei, S. flexneri 2a, 3a and 6 (4)) were estimated.
Serotype-specific sensitivity, specificity, and positive predictive values were calculated for each of the four serotyping methods using the manually adjudicated whole-genome sequencing result as the gold standard. A true positive was defined as a correctly identified serotype by both the method under evaluation and the gold standard. A false positive was identified a serotype incorrectly assigned by the method under evaluation when the gold standard designated it as a different serotype. These misclassifications simultaneously generate a false negative for the true serotype. Additionally, false negatives include “not typed” and “negative” results, yet this applies exclusively to real-time PCR-based methods and reflects their structural inability to detect serotypes for which the algorithm lacks gene targets. Ninety-five percent confidence intervals were estimated using the Wilson score method.
Visualization and statistical analysis were performed in RStudio 2024.04.2.
Results
Among the 1117 participants enrolled in the Peru EFGH study, 228 had a Shigella attributable diarrhea (10.2% (114/1117) culture-based prevalence; 19.2% (214/1117) real-time PCR-based prevalence). For this analysis we excluded children without a Shigella isolate, however serotypes assigned by stool-based real-time PCR are presented in S3 Table. Results of all four serotyping methods, (i) isolate based agglutination, (ii) isolate based real-time PCR serotyping, (iii) stool-based real-time PCR serotyping, and (iv) in silico genome-based serotyping, were available on 107 out of the 114 Shigella isolates (S2 Table; S1 Fig). Whole genome sequencing was successfully performed for 111 out of the 114 available isolates, and isolate based real-time PCR was performed on 110 of the 114 isolates. All four methods identified a similar number of isolates as S. flexneri: agglutination identified 81/114 (71.1%), isolate-based real-time PCR 77/110 (70.0%), stool-based real-time PCR 76/114 (66.7%) and the in silico method 78/111 (70.3%). All methods identified 33 (28.9%) isolates as S. sonnei, except the stool-based real-time PCR which identified 32. The detailed number of S. flexneri serotypes identified by each serotyping method can be visualized in Table 2.
Table 2. Shigella serogroup and Serotyping results by serotyping technique.
| Shigella Serogroup and Serotype | Definitive Serotype | Agglutination (n (%)) | Real-time PCR of Isolate (n (%)) | Real-time PCR of Stool (n (%)) | In-silico (n (%)) |
|---|---|---|---|---|---|
| S. flexneri | 78 (68.4%) | 81 (71.1%) | 77 (67.5%) | 76 (66.7%) | 78 (68.4%) |
| 1a | 9 (11.5%) | 9 (11.1%) | 3 (3.9%) | 3 (3.9%) | 4 (5.1%) |
| 1b | 4 (5.1%) | 4 (4.9%) | 9 (11.7%) | 8 (10.5%) | 9 (11.5%) |
| 2a | 36 (46.2%) | 39 (48.1%) | 36 (46.8%) | 35 (46.1%) | 36 (46.2%) |
| 2b | 12 (15.4%) | 11 (13.6%) | 12 (15.6%) | 11 (14.5%) | 12 (15.4%) |
| 3a | 5 (6.4%) | 5 (6.2%) | 5 (6.5%) | 6 (7.9%) | 6 (7.7%) |
| 4a | 2 (2.6%) | 11 (13.6%) | 2 (2.6%) | 1 (1.3%) | 0 (0%) |
| 4av | 0 (0%) | 0 (0%) | 0 (0%) | 0 (0%) | 2 (2.6%) |
| 4b | 0 (0%) | 0 (0%) | 0 (0%) | 3 (3.9%) | 0 (0%) |
| Y | 3 (3.8%) | 2 (2.5%) | 0 (0%) | 0 (0%) | 2 (2.6%) |
| Yv | 7 (9.0%) | 0 (0%) | 0 (0%) | 0 (0%) | 7 (9.0%) |
| Not Typed | 0 (0%) | 0 (0%) | 10 (13.0%) | 9 (11.8%) | 0 (0%) |
| S. sonnei | 33 (28.9%) | 33 (28.9%) | 33 (28.9%) | 32 (28.1%) | 33 (28.9%) |
| Negative | 0 (0%) | 0 (0%) | 0 (0%) | 6 (5.3%) | 0 (0%) |
| Not Performed | 3 (2.6%) | 0 (0%) | 4 (3.5%) | 0 (0%) | 4 (2.6%) |
| Total | 114 | 114 | 114 | 114 | 114 |
Across all serotyping methods, the main misclassifications were associated with serotype Y (n = 3) and serotype Yv (n = 7). Specifically, agglutination-based serotyping classified all 7 Yv isolates as 4a and one of the three Y isolates also as 4a. The in silico-based prediction tool correctly classified all 7 Yv isolates and 2 of the 3 Y isolates (one was incorrectly classified as a 1a). real-time PCR does not have the ability to serotype Y or Yv isolates. Thus, using isolate-based real-time PCR a total of nine Yv/Y isolates were “non-typed”, while one Y isolate was incorrectly classified as 1a. Similarly, stool-based real-time PCR serotyping provided a “non-typed” result for 5 Yv isolates and 2 Y isolates and incorrectly misclassified 2 Yv isolates as 4b and 1 Y isolate as 1a.
The second most common misclassification was associated with S. flexneri 1a. There were 9 isolates confirmed as 1a which were correctly classified as such by agglutination. However, the other three typing methods misclassified up to 7 of these as 1b. Manual evaluation of the contigs identified that the oac gene was present yet it contained a nonsense mutation. In-silico based serotype prediction misclassified three additional isolates: two S. flexneri 4a isolates were classified as 4av, yet manual examination of the contigs revealed the presence of the Petn-transferase moiety with a nonsense mutation, and one 1b isolate was classified as a 3a isolate, yet the gtrI and oac genes were confirmed to be present.
In rare instances, stool-based real-time PCR could not detect Shigella in stool, presumably due to stool sampling heterogeneity. Specifically, six samples were negative for the ipaH gene, which did not allow for stool-based genotyping. Additionally, in two instances the required cycle threshold was not achieved to call a serotype, obtaining a “not typed’ output on a 1a isolate and on a 2a isolate.
Besides S. sonnei, S. flexneri 2a and 2b showed the highest concordance across all four serotyping methods. These two serotypes were also the most common S. flexneri serotypes in the current data set (2a: 46.2% (36/114); 2b: 15.4% (12/114)). Discrepancies associated with these two serotypes were limited.
Fig 2 compares the results from the four serotyping methods to the gold standard of adjudicated manual sequence analysis, while serotype-specific sensitivity, specificity, and positive predictive value for each method are provided in S4 Table. The accuracy of agglutination was 91.9% (102/111; 95% CI: 85.3% - 97.7%, Wilson), and that of ShigaPass was 91.0% (101/111; 95% CI: 84.2% - 95.0%, Wilson). The accuracy of the real-time PCR method on isolates was 91.8% (90/98; 95% CI: 84.7%-95.8%, Wilson). This calculation excludes excluding the S. flexneri Y and Yv samples (which carry no typable genes) that were accurately “not-typed” by the isolate based real-time PCR (n = 9) When including the “not-typed” as discrepancies the accuracy was 84.1% (90/107; 95% CI: 76.0%-89.8%, Wilson). Similarly, the accuracy of the stool-based real-time PCR was 87.9% (87/99; 95% CI: 80.0%-92.9%, Wilson), when excluding “negative” (n = 5) results and “non-typed” S. flexneri Y and Yv (n = 7) from the analysis. Including these samples decreased accuracy to 78.3% (87/111; 95% CI: 69.8%-85.0%, Wilson). Complete concordance was achieved in 89.2% (83/93) of samples evaluated when excluding non-typed S. flexneri Y and Yv samples and negative outputs from real-time PCR (n = 14), and in 77.6% (83/107) when including them. Table 3 lists discrepancies and indicates the correct serotype assignment after identification of genes and/ or mutations causing discrepancies.
Fig 2. Serotype assignment of agglutination, isolate-based real-time PCR, stool-based real-time PCR and in-Silico in compared to the definitive serotype adjudicated by manual sequence analysis.
Legend: Each panel shows a serotype concordance matrix in which rows represent the assigned serotype, columns represent the definitive serotype, and cell values indicate the number of isolates. Cells colored in pink or red indicate misclassifications, with color intensity proportional to the number of misclassified isolates. Gray cells indicate isolates that were undetected or non-typable by the method.
Table 3. Shigella serotyping discrepancies between culture agglutination, Real-time PCR and in-silico typing, and detection of genes and mutations leading to discrepancies.
| ID | Serogroup | Agglutination | Real-time PCR of Isolate | Real-time PCR of Stool | In-Silico | Definitive Serotype | Correct Assignment Explanation | Platform(s) with Correct Assignment |
|---|---|---|---|---|---|---|---|---|
| Discrepancies due to at least one serotyping mismatch | ||||||||
| 005005 | flexneri | 1a | 1b | 1b | 1b | 1a | oac (1b) gene w/ nonsense mutation | Agglutination |
| 005035 | flexneri | 1a | 1b | 1b | 1b | 1a | oac (1b) gene w/ nonsense mutation | Agglutination |
| 007901 | flexneri | 1a | 1b | 1b | 1b | 1a | oac (1b) gene w/ nonsense mutation | Agglutination |
| 014771 | flexneri | 1a | 1b | 1b | 1b | 1a | oac (1b) gene w/ nonsense mutation | Agglutination |
| 014681 | flexneri | 1a | 1b | 1b | 1b | 1a | oac (1b) gene w/ nonsense mutation | Agglutination |
| 014741 | flexneri | 1a | 1b | 1b | 1b | 1a | oac (1b) gene w/ nonsense mutation | Agglutination |
| 012476 | flexneri | 1a | NA | Not typed | 1a | 1a | gtrI gene present | Agglutination/ In-silico |
| 012103 | flexneri | 1b | 1b | 3a | 3a | 1b | gtrI and oac gene present | Agglutination/ real-time PCR of Isolate/ In-silico |
| 014581 | flexneri | 2a | 2b | Negative | 2b | 2b | oacD gene with frameshift | real-time PCR / In-silico |
| 014625 | flexneri | 2a | 2a | Not typed | 2a | 2a | gtrII gene present | Agglutination/ real-time PCR of Isolate/ In-silico |
| 002187 | flexneri | 4a | Not typed | Not typed | Yv | Yv | No gtr gene. Petn-transferase moiety present | In-silico |
| 006825 | flexneri | 4a | Not typed | Not typed | Yv | Yv | No gtr gene. Petn-transferase moiety present | In-silico |
| 006961 | flexneri | 4a | Not typed | Not typed | Yv | Yv | No gtr gene. Petn-transferase moiety present | In-silico |
| 007841 | flexneri | 4a | 4a | 4a | 4av | 4a | gtrIV gene present. Petn-transferase moiety present with a nonsense mutation | Agglutination/ real-time PCR |
| 011149 | flexneri | 4a | Not typed | Not typed | Yv | Yv | No gtr gene. Petn-transferase moiety present | In-silico |
| 014589 | flexneri | 4a | 4a | Negative | 4av | 4a | Has Petn-transferase but lpt-O gene it has a nonsense mutation | Agglutination/ real-time PCR of Isolate |
| 014593 | flexneri | 4a | Not typed | 4b | Yv | Yv | No gtr gene. Petn-transferase moiety present | In-silico |
| 014601 | flexneri | 4a | Not typed | 4b | Yv | Yv | No gtr gene. Petn-transferase moiety present | In-silico |
| 014633 | flexneri | 4a | Not typed | Not typed | Yv | Yv | No gtr gene. Petn-transferase moiety present | In-silico |
| 014697 | flexneri | 4a | Not typed | Not typed | Y | Y | No gtr gene | In-silico |
| 011803 | flexneri | Y | 1a | 1a | 1a | Y | No gtr gene | Agglutination |
| 000840 | flexneri | 4a | Not typed | 4b | NA | NA | Not available | |
| 001671 | flexneri | 2a | 2a | Negative | NA | NA | Not available | |
| Discrepancies due to negatives and non-typable in stool samples only | ||||||||
| 009922 | flexneri | 1b | 1b | Negative | 1b | 1b | gtrI and oac gene present | Agglutination/ real-time PCR of Isolate/ In-silico |
| 014749 | flexneri | 2a | 2a | Negative | 2a | 2a | gtrII gene present | Agglutination/ real-time PCR of Isolate/ In-silico |
| 008864 | flexneri | Y | Not typed | Not typed | Y | Y | No gtr gene | Agglutination/ In-silico |
| 014163 | sonnei | sonnei | sonnei | Negative | sonnei | sonnei | S. sonnei | Agglutination/ real-time PCR of Isolate/ In-silico |
NA = Not Available.
Finally, the accuracy of all serotyping methods exceeded 90% for all three vaccine combinations. Agglutination, isolate-based real-time PCR and in silico methods demonstrated almost perfect agreement for the serotypes included in these products (Table 4).
Table 4. Accuracy of all four serotyping methods for Shigella serogroups and serotypes included in current Shigella licensed vaccine and major vaccine candidates.
| Serogroup and Serotype combination | Number of Samples Fully Evaluated | Accuracy (% (n/N); 95% CI, Wilson) | |||
|---|---|---|---|---|---|
| Agglutination | Isolate-based real-time PCR | Stool-based real-time PCR | In-silico | ||
| Candidate A (S. sonnei, S. flexneri 1b, 2a and 3a) | 75 | 100% (75/75); 95.0-100% | 100 (75/75); 95.0-100% | 93.3% (70/75); 85.3%-97.1% | 98.7 (74/75); 92.8%-99.8% |
| Candidate B (S. sonnei; S. flexneri 2a) | 67 | 100% (67/67); 94.6-100% | 100 (67/67); 94.6-100% | 95.6% (64/67); 87.6%-98.5% | 100 (67/67); 94.6-100% |
| Candidate C (S. sonnei, S. flexneri 2a, 3a and 6) | 39 | 100% (39/39); 91.0-100% | 100 (39/39); 91.0-100% | 94.9% (37/39); 83.1%-98.6% | 100 (39/39); 91.0-100% |
Discussion
Commercial antisera for serogrouping (Shigella dysenteriae, Shigella flexneri, Shigella boydii, and Shigella sonnei) and then serotyping of the isolates into serotypes involves agglutination, a test that has minimal logistical requirements and can be performed at basic regional laboratories [28,29]. However, there are limitations of agglutination-based serotyping, which is costly and time-consuming, as well as harder to standardize across different laboratories [29,30]. Additionally, as shown in the EFGH study, culture sensitivity compared to real-time PCR is considerably lower, observed agglutination reactions are also somewhat subjective, and cross-reactivity between serotypes is also possible. Indeed, we observed that antisera against S. flexneri 4a reacted to S. flexneri Yv strains that did not possess the gtrIV encoding the 1,6 glucosyltransferase that creates the 4a epitope [13]. More importantly, there is no international standard and the companies producing antisera often discontinue production or do not possess antisera to newly discovered serotypes. Such is the case of serotypes Yv, 4av and Xv. These serotypes originate by the modification of the O-antigen with phosphoethanolamine (PEtN), which is encoded by the plasmid borne gene opt [7,12]. The difference in these serotypes is the position on the rhamnose that is substituted with PEtN. In both serotype Yv and 4a, the modification of the O-antigen is given at position 3 of rhamnose chain (RhaIII), predominantly. In serotype Xv (not observed in this study), the O-antigen is also modified with PEtN, but on RhaII [7,12,31]. The differentiated pattern in phosphorylation is attributed to a difference of 11 bases (and 7 associated amino acid changes) in the gene sequences of the lpt-O gene homolog that is found in serotypes Yv and 4av (opt) versus serotype Xv (lpt-ORII). The availability of antisera to correctly type and identify these serotypes (Reagensia AB, Sweden) is no longer marketed, causing an incomplete categorization and profiling of Shigella serotypes that appears to have no clear solution. In the current study, the misidentification of serotype Yv as 4a was the most common serotyping error identified. Additionally, inconsistent typing results using different commercial and reference antisera is a well-documented phenomenon [16]. These shortcomings speak to the value of moving to real-time PCR or sequence-based methodologies as more reproducible and rigorous over time and place, as the same isolate will consistently be able to be fully typed over time independent of market forces.
A practical alternative to serotyping based on agglutination is either a real-time PCR-based typing strategies or in silico analysis of whole genome sequencing of culture derived isolates. Both real-time PCR-based and in silico analysis classify serotypes according to rfb genes cluster sequences, bacteriophage seroconverting genes encoding O-antigen glycosylation and/ or O-acetylation enzymes that modify the O-antigen chain at specific positions [6,8,9,11,13], and plasmid encoded modification [7,12]. Real-time PCR genotyping strategies, such as the real-time PCR serotyping strategy used here, have the advantage of a quick, single step serotype identification and has been found by other groups to have improved accuracy relative to agglutination when done on isolates [32]. For state laboratories or vaccine trials, this method is the most efficiently to scale, limit costs and time spent typing isolates, as well as limit the cost associated with whole genome sequencing and associated expert requirements for bioinformatic sequencing. In the present analysis, isolate based real-time PCR serotyping performed similarly to agglutination and whole genome sequencing. It is limited, however, by its inability to detect Y and Yv serotypes. Although these serotypes have relatively low prevalence in most settings, the application of qPCR-based serotyping would be restricted in geographic areas where they are more common. The current real-time PCR algorithm is not designed to detect the Shigella O-antigen structure without any structural modification, as well as the absence of the opt gene that codes for PEtN from the PCR mixture. Additionally, it failed in the detection of a nonsense mutation in the oac gene reverting a 1b into a 1a serotype. However, this tool has the advantage of being customizable and adaptable to new Shigella serotypes and could potentially be improved to overcome current misclassifications. Moreover, the stool-based real-time PCR serotyping method is advantageous because it can serotype Shigella straight from fecal samples without the need for culturing [23,33]. Additionally, it shows approximately two-fold higher sensitivity relative to culture which may provide more robust estimates of serotype distribution, particularly in remote settings where specimen transport and culture may by logistically limited [34]. Although its accuracy may be marginally lower than that of the three other serotyping strategies, this limitation is offset by the increased sensitivity of real-time PCR to detect Shigella in stool compared with culture. This would allow completion of a vaccine efficacy trial with fewer participants.
The WGS in silico serotyping tool utilized by this study, ShigaPass, presented certain limitations which are not commonly recognized. ShigaPass requires the complete genome assembly as input, including contigs below standard-length thresholds. Our default assembly pipeline applied a 500 nt minimum contig cutoff; however, for one isolate this excluded a contig carrying a fragment of ipaH, reducing the gene representation below the minimum kmer threshold required for ShigaPass identification. Lowering the contig cutoff to include contigs below 500 nt for this isolate recovered the ipaH-containing contig and enabled successful species assignment. Second, ShigaPass does not detect nonsense mutations that make genes non-functional, and thus revert the serotype being encoded. For instance, ShigaPass identified two samples as serotype 4av, failing to identify a nonsense mutation in the opt gene that encodes for PEtN that would revert it to serotypes to 4a. Additionally, ShigaPass also identified 7 samples as S. flexneri seroptype 1b, failing to identify a nonsense mutation in the oac gene, reverting it to serotype 1a.
When described in the literature, a correct assignment of serotype among 4879 genomes was observed in 98.5% and was noted to perform particularly well among Shigella boydii and Shigella dysentariae serogroups [27]. Other in silico typing tools exist (ShigaTyper [35] and ShigEiFinder [36]), but are generally considered less accurate than ShigaPass [27]. The differences between those attained by Yassine et al and the current analysis could be a function of the way genomes were selected. Yassine et al selected as reference strains from as many serotypes as available from all serogroups. The current analysis tested samples selected from a study that sampled children between the ages of 6–36 months presenting with diarrhea in a LMIC setting like one that would be considered for a vaccine trial. The relative distribution therefore reflects real world performance. In the current analysis, no Shigella dysenteriae was seen in this population, nor was there any S. boydii. The outcome metrics are thus driven by the distribution of serotypes in the target population for vaccine development and driven nearly entirely by the performance of the typing schemes for S. flexneri serogroups (all strategies performed well with S. sonnei) [27].
WGS provides the advantage of further genomic analysis. As mentioned above, we demonstrated that WGS reads could be mapped to the O-antigen biosynthesis and modification genes identifying nonsense mutations that would lead to eliminate the function of a gene that was scored as present by ShigaPass. Moreover, additional analysis provided within EnteroBase such as cgMLST, SNP calls, and antimicrobial resistance (AMR) markers can improve the understanding of the epidemiology of shigellosis, including phylogenomic investigations and prevalence and expansion of AMR. Although sequencing and bioinformatic analyses may involve higher costs and demand greater technical expertise, integrating an in silico tool such as ShigaPass into a web-based platform like EnteroBase or PubMLST could help overcome these costs, particularly as the cost of whole-genome sequencing continues to decline. That said, the relative increased cost, processing time, and bioinformatics restrict the utility of the widespread implementation of WGS for typing.
Vaccines under development are currently largely multivalent [37]. Calculating serotype specific protection, and possible cross protection among antigenically related serogroups to those contained by the vaccine require accurate serotyping strategies. Considering the current vaccine candidates in advanced stages of development [2–4], performance of all serotyping techniques is considered high and could be considered adequate for evaluating serotype specific protection during clinical trials.
The present study is limited by the restricted number of serotypes found in Loreto, Peru, as well as the geographic restriction of the population sampled. Thus, serotyping performance of other serotypes not found in this region could not be determined. That said, the serotypes found in this population are representative of the most frequently identified in large scale epidemiologic studies [38].
In conclusion, in silico tools such as ShigaPass provide a valuable step for serotyping S. flexneri genomes when complemented by mapping of DNA sequence reads to the O-antigen biosynthesis and modification genes of interest. Together these steps helped correctly serotype Shigella isolates in the absence of commercial antisera that was unavailable to perform appropriate and complete culture agglutination of isolates. We also note that real-time PCR-based serotyping is a valuable strategy that can be expanded to new serogroups with additional primer construction and can inform serotyping in large-scale studies that cannot include culture and subsequent whole genome sequence Shigella spp.
Financial support. Funding for this study was provided by the National Institutes of Health of the United States (RO1AI158576 and R21AI163801 to MNK and CTP, D43TW010913 to MNK and MPO; K43TW012298 to FS). The Gates Foundation to PP (INV-016650 and INV-131791) and EH (INV-028721 and INV-041730) provided additional support to the EFGH study. MNK also received funding from the University of Virginia. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Supporting information
(DOCX)
(DOCX)
(DOCX)
(DOCX)
(DOCX)
Acknowledgments
We would like to thank Dr. Nancy Strockbine for inputs on the early analysis and development of this work.
Data Availability
All data generated or analyzed during this study are included in this published article. All genome sequences generated in this study have been deposited in the NCBI database under BioProject accession number PRJNA1321318.
Funding Statement
Funding for this study was provided by the National Institutes of Health of the United States (RO1AI158576 and R21AI163801 to MNK and CTP, D43TW010913 to MNK and MPO; K43TW012298 to FS). The Gates Foundation to PP (INV-016650 and INV-131791) and EH (INV-028721 and INV-041730) provided additional support to the EFGH study. MNK also received funding from the University of Virginia. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
References
- 1.Collaborators GBDDD. Global, regional, and national age-sex-specific burden of diarrhoeal diseases, their risk factors, and aetiologies, 1990-2021, for 204 countries and territories: a systematic analysis for the Global Burden of Disease Study 2021. Lancet Infect Dis. 2025;25(5):519–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Leroux-Roels I, Maes C, Mancini F, Jacobs B, Sarakinou E, Alhatemi A, et al. Safety and Immunogenicity of a 4-Component Generalized Modules for Membrane Antigens Shigella Vaccine in Healthy European Adults: Randomized, Phase 1/2 Study. J Infect Dis. 2024;230(4):e971–84. doi: 10.1093/infdis/jiae273 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Mo Y, Fang W, Li H, Chen J, Hu X, Wang B, et al. Safety and Immunogenicity of a Shigella Bivalent Conjugate Vaccine (ZF0901) in 3-Month- to 5-Year-Old Children in China. Vaccines (Basel). 2021;10(1):33. doi: 10.3390/vaccines10010033 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Kelly M, Janardhanan J, Wagh C, Verma S, Charles RC, Leung DT, et al. Development of a Shigella conjugate vaccine targeting Shigella flexneri 6 that is immunogenic and provides protection against virulent challenge. Vaccine. 2024;42(24):126263. doi: 10.1016/j.vaccine.2024.126263 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.MacLennan CA, Grow S, Ma L-F, Steele AD. The Shigella Vaccines Pipeline. Vaccines (Basel). 2022;10(9):1376. doi: 10.3390/vaccines10091376 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Allison GE, Verma NK. Serotype-converting bacteriophages and O-antigen modification in Shigella flexneri. Trends Microbiol. 2000;8(1):17–23. doi: 10.1016/s0966-842x(99)01646-7 [DOI] [PubMed] [Google Scholar]
- 7.Sun Q, Lan R, Wang J, Xia S, Wang Y, Wang Y, et al. Identification and characterization of a novel Shigella flexneri serotype Yv in China. PLoS One. 2013;8(7):e70238. doi: 10.1371/journal.pone.0070238 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Sun Q, Lan R, Wang Y, Wang J, Wang Y, Li P, et al. Isolation and genomic characterization of SfI, a serotype-converting bacteriophage of Shigella flexneri. BMC Microbiol. 2013;13:39. doi: 10.1186/1471-2180-13-39 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Verma NK, Brandt JM, Verma DJ, Lindberg AA. Molecular characterization of the O-acetyl transferase gene of converting bacteriophage SF6 that adds group antigen 6 to Shigella flexneri. Mol Microbiol. 1991;5(1):71–5. doi: 10.1111/j.1365-2958.1991.tb01827.x [DOI] [PubMed] [Google Scholar]
- 10.Adhikari P, Allison G, Whittle B, Verma NK. Serotype 1a O-antigen modification: molecular characterization of the genes involved and their novel organization in the Shigella flexneri chromosome. J Bacteriol. 1999;181(15):4711–8. doi: 10.1128/JB.181.15.4711-4718.1999 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Mavris M, Manning PA, Morona R. Mechanism of bacteriophage SfII-mediated serotype conversion in Shigella flexneri. Mol Microbiol. 1997;26(5):939–50. doi: 10.1046/j.1365-2958.1997.6301997.x [DOI] [PubMed] [Google Scholar]
- 12.Sun Q, Knirel YA, Lan R, Wang J, Senchenkova SN, Jin D, et al. A novel plasmid-encoded serotype conversion mechanism through addition of phosphoethanolamine to the O-antigen of Shigella flexneri. PLoS One. 2012;7(9):e46095. doi: 10.1371/journal.pone.0046095 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Adams MM, Allison GE, Verma NK. Type IV O antigen modification genes in the genome of Shigella flexneri NCTC 8296. Microbiology (Reading). 2001;147(Pt 4):851–60. doi: 10.1099/00221287-147-4-851 [DOI] [PubMed] [Google Scholar]
- 14.Levine MM, Kotloff KL, Barry EM, Pasetti MF, Sztein MB. Clinical trials of Shigella vaccines: two steps forward and one step back on a long, hard road. Nat Rev Microbiol. 2007;5(7):540–53. doi: 10.1038/nrmicro1662 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Liu J, Pholwat S, Zhang J, Taniuchi M, Haque R, Alam M, et al. Evaluation of Molecular Serotyping Assays for Shigella flexneri Directly on Stool Samples. J Clin Microbiol. 2021;59(2):e02455-20. doi: 10.1128/JCM.02455-20 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Brengi SP, Sun Q, Bolaños H, Duarte F, Jenkins C, Pichel M, et al. PCR-Based Method for Shigella flexneri Serotyping: International Multicenter Validation. J Clin Microbiol. 2019;57(4):e01592-18. doi: 10.1128/JCM.01592-18 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Sun Q, Lan R, Wang Y, Zhao A, Zhang S, Wang J, et al. Development of a multiplex PCR assay targeting O-antigen modification genes for molecular serotyping of Shigella flexneri. J Clin Microbiol. 2011;49(11):3766–70. doi: 10.1128/JCM.01259-11 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Atlas HE, Conteh B, Islam MT, Jere KC, Omore R, Sanogo D, et al. Diarrhea Case Surveillance in the Enterics for Global Health Shigella Surveillance Study: Epidemiologic Methods. Open Forum Infect Dis. 2024;11(Suppl 1):S6–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Yousafzai MT, Cornick J, Yori PP, Hossain MJ, Keita AM, Hannah EA, et al. Incidence and antimicrobial resistance of Shigella-attributable diarrhea in young children: Results from the multi-country enterics for global health (EFGH) Shigella Surveillance study. 2025.
- 20.Manzanares Villanueva K, Pinedo Vasquez T, Peñataro Yori P, Romaina Cacique L, Garcia Bardales PF, Shapiama Lopez WV, et al. The Enterics for Global Health (EFGH) Shigella Surveillance Study in Peru. Open Forum Infect Dis. 2024;11(Suppl 1):S121–8. doi: 10.1093/ofid/ofad655 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Horne B, Badji H, Bhuiyan MTR, Romaina Cachique L, Cornick J, Hotwani A, et al. Microbiological Methods Used in the Enterics for Global Health Shigella Surveillance Study. Open Forum Infect Dis. 2024;11(Suppl 1):S25–33. doi: 10.1093/ofid/ofad576 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Rogawski McQuade ET, Liu J, Mahfuz M, Havt A, Varghese T, Shrestha J, et al. Epidemiology of Shigella species and serotypes in children: a retrospective substudy of the MAL-ED observational birth cohort study. Lancet Microbe. 2025;6(6):101064. doi: 10.1016/j.lanmic.2024.101064 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Liu J, Garcia Bardales PF, Islam K, Jarju S, Juma J, Mhango C, et al. Shigella Detection and Molecular Serotyping With a Customized TaqMan Array Card in the Enterics for Global Health (EFGH): Shigella Surveillance Study. Open Forum Infect Dis. 2024;11(Suppl 1):S34–40. doi: 10.1093/ofid/ofad574 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Schiaffino F, Parker CT, Paredes Olortegui M, Pascoe B, Manzanares Villanueva K, Garcia Bardales PF, et al. Genomic resistant determinants of multidrug-resistant Campylobacter spp. isolates in Peru. J Glob Antimicrob Resist. 2024;36:309–18. doi: 10.1016/j.jgar.2024.01.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Zhou Z, Alikhan N-F, Mohamed K, Fan Y, Agama Study Group, Achtman M. The EnteroBase user’s guide, with case studies on Salmonella transmissions, Yersinia pestis phylogeny, and Escherichia core genomic diversity. Genome Res. 2020;30(1):138–52. doi: 10.1101/gr.251678.119 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Parks DH, Imelfort M, Skennerton CT, Hugenholtz P, Tyson GW. CheckM: assessing the quality of microbial genomes recovered from isolates, single cells, and metagenomes. Genome Res. 2015;25(7):1043–55. doi: 10.1101/gr.186072.114 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Yassine I, Hansen EE, Lefèvre S, Ruckly C, Carle I, Lejay-Collin M, et al. ShigaPass: an in silico tool predicting Shigella serotypes from whole-genome sequencing assemblies. Microb Genom. 2023;9(3):mgen000961. doi: 10.1099/mgen.0.000961 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Bopp CA, Ries AA, Wells J. Laboratory methods for the diagnosis of epidemic dysentery and cholera. 1999.
- 29.Kaminski RW, Pavlinac PB, Platts-Mills JA, Rogawski McQuade ET, Hausdorff WP, Isbrucker RA, et al. WHO Workshop Report: Regulatory Science to Inform Clinical Pathways for Shigella Vaccines Intended for Use in Children in Low- and Middle-Income Countries. Vaccines [Internet]. 2025;13(5):439. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Bhuiyan TR, Liu J, Juma J, Horne BA, Hotwani A, Badji H, et al. Optimizing Shigella isolation: A multi-site evaluation of laboratory culture methods for Shigella detection, speciation, and serotyping with different transport media and sample types in the Enterics for Global Health study. medRxiv. 2025. doi: 2025.09.02.25334652 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Knirel YA, Lan R, Senchenkova SN, Wang J, Shashkov AS, Wang Y, et al. O-antigen structure of Shigella flexneri serotype Yv and effect of the lpt-O gene variation on phosphoethanolamine modification of S. flexneri O-antigens. Glycobiology. 2013;23(4):475–85. doi: 10.1093/glycob/cws222 [DOI] [PubMed] [Google Scholar]
- 32.Impact of malnutrition on small intestinal repair after acute viral enteritis. Nutr Rev. 1985;43(11):335–7. doi: 10.1111/j.1753-4887.1985.tb02394.x [DOI] [PubMed] [Google Scholar]
- 33.Liu J, Gratz J, Amour C, Kibiki G, Becker S, Janaki L, et al. A laboratory-developed TaqMan Array Card for simultaneous detection of 19 enteropathogens. J Clin Microbiol. 2013;51(2):472–80. doi: 10.1128/JCM.02658-12 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Liu J, Platts-Mills JA, Juma J, Kabir F, Nkeze J, Okoi C, et al. Use of quantitative molecular diagnostic methods to identify causes of diarrhoea in children: a reanalysis of the GEMS case-control study. Lancet. 2016;388(10051):1291–301. doi: 10.1016/S0140-6736(16)31529-X [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Wu Y, Lau HK, Lee T, Lau DK, Payne J. In silico serotyping based on whole-genome sequencing improves the accuracy of Shigella identification. Appl Environ Microbiol. 2019;85(7). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Zhang X, Payne M, Nguyen T, Kaur S, Lan R. Cluster-specific gene markers enhance Shigella and enteroinvasive Escherichia coli in silico serotyping. Microb Genom. 2021;7(12):000704. doi: 10.1099/mgen.0.000704 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Martin P, Alaimo C. The Ongoing Journey of a Shigella Bioconjugate Vaccine. Vaccines (Basel). 2022;10(2):212. doi: 10.3390/vaccines10020212 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Livio S, Strockbine NA, Panchalingam S, Tennant SM, Barry EM, Marohn ME, et al. Shigella isolates from the global enteric multicenter study inform vaccine development. Clin Infect Dis. 2014;59(7):933–41. doi: 10.1093/cid/ciu468 [DOI] [PMC free article] [PubMed] [Google Scholar]