Skip to main content
Premium Trial:

Request an Annual Quote

Web-Based Phylogenetic Tools Offer Insights Into Viral Evolution, Geographic Spread

Premium

NEW YORK (GenomeWeb) – A pair of researchers from the Max Planck Institute of Developmental Biology and Fred Hutchinson Cancer Research Center who have developed a web-based application for tracking the evolution of the seasonal influenza virus have now created versions of their tool to help researchers track the evolution of two other troublesome viruses — the Ebola virus and the Middle East respiratory syndrome coronavirus (MERS-CoV).

The scientists developed the original Nextflu application, which came out earlier this year, to help developers of seasonal flu vaccines make use of viral sequencing and other data collected by laboratories around the world. As described in a recent Bioinformatics paper, Nextflu displays the genetic and epidemiological relationships that exist between influenza viruses. It uses a python-based processing pipeline to clean and filter input viral sequences from recent influenza strains and a visualization tool that displays the processed information.

"There's been a lot of interest in flu from an evolutionary biology perspective ... and a lot of work on studying it," Trevor Bedford, a researcher in the infectious disease division of the Fred Hutchinson Cancer Research Center, told GenomeWeb in a recent conversation. "What we were trying to do with Nextflu was get a robust pipeline in place that could do these sorts of evolutionary analyses and present them to people who were interested [in flu]." 

Nextflu uses data from the EpiFlu database, a repository of influenza sequences and clinical and epidemiological data that is maintained by the Global Initiative on Sharing Avian Influenza Data, an international consortium that enables and promotes the sharing of data from various influenza virus types. Viruses added to Nextflu are displayed in a colorful phylogenetic tree that shows the viral strains circulating today and the relationships between them. 

Furthermore, strains are annotated with information such as viral genotype at specific sites, and sampling location. Other tools estimate mutation, genotype, and clade frequency trajectories which help with predictions about future viral activity — this information is also shown in the tree. Users can filter which strains they see in the tree according to mutations associated with antibody binding, by local branching indices, by genotype at specific amino acid positions, or by geographic region. Researchers can view the evolution of about 1,000 viruses at a time shown over a three-year window, or they can look at the same number of viruses over the course of a year, Bedford said. The latter option provides a more deeply-sampled picture.

Nextflu displays data from four current flu viruses that go into the annual flu vaccine: H3N2 and H1N1, which are both subtypes of influenza A; and B-Victoria and B-Yamagata, both subtypes of influenza B. So far, Bedford said, the tool is being used by some researchers working in World Health Organization collaborating centers. "It's I think the best way for the different research groups to see each other's data," he said. With Nextflu, "we get this global picture of things which I think is useful."

However there's nothing in the underlying code and pipeline for Nextflu that restrict its use to influenza. The developers are already looking into creating versions of the tool focused on a variety of viruses. So far, they have constructed and released this summer an iteration that showcases the evolutionary relationships between viral strains from the recent Ebola outbreak in West Africa and a second iteration showing relationships between strains of MERS-CoV

Bedford and his co-authors have made their code available on Github for interested researchers to download and use to explore aspects of viral evolution that are of interest to them. Moving forward, the developers are working on making the original Nextflu infrastructure even more generalizable than it already is so that it can be adapted to work for even more viruses. It's pretty flexible right now, according to Bedford, although there was some software engineering and some parameter tweaking required to move from influenza to other viruses. The next step is to "make something that's more easily generalizable to different viruses and ...  [make it easy] to stand-up a version for a particular outbreak or particular virus of interest," he said.

They are also doing some work on the influenza version of the application. "There's a lot of excitement about predictive models — so taking some of these viral characteristics and trying to do a full statistical model that predicts, [based on] the strains circulating in the world today, what we think the strain make-up in the world will look like a year from now," Bedford said. Retrospective studies have shown that these sorts of models actually work quite well for predicting flu strain evolution and "so we are trying to get those models working in Nextflu as well," he said.