The ArboTyping tool: monitoring virus genotypes to track disease outbreaks
There is a growing need to improve prevention and control strategies to tackle the increasingly frequent and intense outbreaks of vector-borne diseases worldwide. Monitoring of virus genotype diversity is an important aspect of tracking the emergence and evolution of these outbreaks and, in recent work looking at dengue, Zika and chikungunya outbreaks, researchers have developed a method to classify virus sequences based on their species and sub-species (i.e. serotype and/or genotype). This ArboTyping tool comes as an easy-to-use software to classify viruses based on whole-genome and partial-genome sequences. In this Infectious Thoughts interview, we speak to Pr. Tulio de Oliveira, from the KwaZulu-Natal Research Innovation and Sequencing Platform (KRISP) based at the University of KwaZuluNatal in Durban, South Africa, and to Pr. Luiz C. J. Alcantara from Fundação Oswaldo Cruz and Universidade Federal de Minas Gerais, Brazil, about the challenges which the ArboTyping tool is tackling and this tool's essential role in improving epidemiological, clinical, and entomological studies as well as diagnostics development.
What are some of the main difficulties in the real-time monitoring of outbreaks and surveillance of diseases caused by arboviruses, which include dengue fever, West Nile fever, yellow fever, Zika? What are the specific gaps in monitoring which you sought to tackle using your innovative approach?
There are many difficulties in real-time monitoring of outbreaks, including: a. Lack or sub-optimal vector surveillance in areas of higher concentrations of reported cases in humans. The reason for that is because most of the arboviruses outbreaks happen in developing countries, which normally lack detailed genomics surveillance in the most affected areas; b. Lack of active surveillance in animal reservoirs; c. Under-reporting of cases of co-infections in humans; d. The absence of a serum bank in public health laboratories in affected countries, including Brazil. A temporal genomic / epidemiological surveillance requires the study of samples collected at different times from different locations. And this is lacking in Brazil; e. Too many gaps in the online filling of notification reports.
2. What are the main attributes of the ArboTyping tool, and what were the main steps in its development?
The main advantage of the ArboTyping tool is that it can accurately detect in seconds the species and genotypes/serotypes of the most common arboviruses. This tool has been instrumental for the identification of outbreaks of Zika (a beta version tool is cited in the Science paper of Farias et al. 2017 that describes the Zika outbreak in Brazil). The tools is commonly used by the Brazilian Minister of Health and PAHO to identify different outbreaks.
The main steps in its development included: a. Data mining and analysis of thousands of virus sequences in a public databases. b. Filtering the sequences with known genotypes to construct a set of reference sequences representing each genotype of that virus. c. Evaluation of the accuracy of the tool (i.e. sensitivity and specificity) and fine tuning until the tool could accurately identify all of the main genotypes of Zika, dengue and Chikungunya.
The tool is really easy to use and will take seconds to analyze a DNA sequence. The process involves users submitting sequences of arbovirus to the tool, identification of virus species by similarity search, creation of an alignment with this set of reference sequences and production of a phylogenetic analysis of the sequence to determine which genotype of the sequence. It is important to note that up to 2000 sequences can be submitted in a unique section, allowing the tool to be useful for large surveillance programs and to annotate sequences in genetic databases.
How can the AbroTyping tool specifically assist the professionals tackling dengue fever?
The major aim of this tool is genotyping to establish specific details regarding viral sequences such as serotype and genotypes. The tool also allow quick identification of the origin of the virus, which are required in modern arboviral surveillance besides mere identification of the pathogen. For example, in the case of dengue virus, many epidemiologists are demanding the determination of genotype and potential geographical origin of the emergent virus. Together, this information facilitates the identification of how outbreaks emerged and dispersed. For example, the tool has been used to identify new outbreaks and/or clinical outcomes, such as the emergence of DENV-2 SEA genotype in the Americas associated with severe dengue epidemics, the emergence of DENV-3 Indian subcontinent genotype in the Americas associated with high incidence and dispersal and the two DENV-1 outbreaks in Hawaii (2001 and 2015) caused by introduction of 2 different genotypes, etc. In the case of CHIKV, the identification of the ECSA genotype in Brazil raised a public health alarm with the concern of the epidemic potential of this new genotype introduction in the region. Similarly, much concern has been raised by arbovirologists and epidemiologists on the emergence of the ZIKV African genotype and the potential threat to public health which shows how invaluable genotyping information is for diagnostics as well for consequent epidemiological, clinical, and entomological studies.
Outline of the classification procedure
Firstly (A), the viral species is determined using BLAST (basic local alignment search tool). When the submitted sequence is a Zika virus, a Neighbor joining tree is constructed to determine the Zika genotype (B). When the submitted sequence is a Chikungunya virus, a Neighbor joining tree is constructed to determine the Chikungunya genotype (C). When the submitted sequence is a Dengue virus, the serotype is determined using another BLAST invocation (D). Based on the inferred serotype, a serotype specific Neighbor joining tree is constructed to determine the Dengue genotype (E, F, G, H).
How easy will it be to integrate the ArboTyping tool into current disease prevention, surveillance and control programmes? What could facilitate or accelerate its adoption?
It is very easy to integrate the tool into current disease prevention, surveillance and control programmes. As previously mentioned, the tool is free and accepts up to 2000 sequences in a unique section. As described in the paper in PLoS NTD, our tool can accurately genotype sequences with fragments as small as 150bp. As for example, recently Hill and collaborators (https://www.biorxiv.org/content/10.1101/520437v1) identified the introduction of the Zika Virus genotype in Angola using our tool. Our tool allowed quick identification of the African genotype as predominant in this recent outbreak.
What might be next steps or further improvements which you would like to work on in future, and which technologies or partnerships would you like to see developed in order to further your goals?
It is important to note that the development of the tool was done in close collaboration with many of the top arbovirus researchers in the world. For example, there is researchers from 24 different organizations that co-authored the tool, including the CDC, Oxford, FioCruz and other key organizations that are working on the surveillance of arboviruses. We have identified a few aspects that we would like to further develop in the tool and we would like to invite researchers and public health officials to work with us in the improvement of our tools.
a. Add new arboviruses to the phylogenetic classification. For example, we recently created a Yellow Fever version of the tool (Faria et al Science 2018) and would like to expand to other arboviruses that are circulating in the developing world.
b. We would like to have an easy to use metagenomic assay that uses the technology of the minion sequencer of nanopore so data could be generated data in real time in the field without the need of specific viruses amplicons. We are current working on the surveillance of arboviruses in Brazil and other countries of Latin America. All this in partnership with the Brazilian Ministry of Health and Pan American Health Organization.
Access the ArboTyping software online here: http://krisp.org.za/tools.php
Link to the full publication for further information:
A computational method for the identification of Dengue, Zika and Chikungunya virus species and genotypes Vagner Fonseca , Pieter J. K. Libin , Kristof Theys , Nuno R. Faria, Marcio R. T. Nunes, Maria I. Restovic, Murilo Freire, Marta Giovanetti, Lize Cuypers, Ann Nowé, Ana Abecasis, Koen Deforche, Gilberto A. Santiago, Isadora C. de Siqueira, Emmanuel J. San, Kaliane C. B. Machado, Vasco Azevedo, Ana Maria Bispo-de Filippis, Rivaldo Venâncio da Cunha, Oliver G. Pybus, Anne-Mieke Vandamme, Luiz C. J. Alcantara , Tulio de Oliveira