A data-driven discovery blossoms from unexpected collaboration

MIS professor lends her data-wrangling skills to bird flu vaccine research
VIrus graphic

Takeaways

  • Data analytics is changing the way virologists study and develop vaccines.
  • Whether its management decisions or virus mutations, the same analytical tools can be used to address questions, spot patterns and help identify solutions.
  • Constructing research teams with different perspectives and interdisciplinary expertise can lead to better results — even unexpected discoveries.

As an assistant professor of management information systems, Carolina Alves de Lima Salge studies social networks, chatbots and computer algorithms in her research. Wrangling big data sets is the name of the game in her world.

But her expertise as a data scientist led to a research collaboration she never imagined when she chose her field – helping virologists who study the avian flu virus conduct research on vaccine development.

The H5N1 influenza can be a highly pathogenic flu strain, commonly called bird flu. According to the Georgia Department of Public Health, it infects waterfowl and can decimate poultry farms, if introduced. The Centers for Disease Control and Prevention has identified sporadic infections in humans, and epidemiologists are monitoring the virus for any evidence it is evolving to spread from person to person.

Birds of a feather crunch data together

It was just by chance that Salge’s close friend, Miriã Ferreira Criado, worked as a research scientist at the USDA Southeast Poultry Research Laboratory in Athens, where she studies vaccine development for H5N1 influenza. Criado reached out to Salge when their tests of vaccines and models generated an amount of data they found unmanageable.

“The coolest thing about this project is that I was just the data scientist,” Salge said. “I knew nothing about the underlying theory, but I was analyzing the data and identifying patterns. It was kind of amazing to see that my analyses ended up yielding findings that were novel, theoretical, and relevant, especially about a topic I knew very little about. It was a data-driven discovery that turned out to be supported by theory.”

Salge’s work unlocked a key part of the research that was published in the American Society of Microbiology’s Journal of Virology. The study highlighted the diversity of virus strains, the complexity of immune responses to the virus and the efficacy of the existing vaccines. It also suggested specific target sites for future vaccines for poultry and possibly humans.

There are currently several vaccines in use to combat the many variants of H5N1. However, no vaccine strain has proven effective against all the influenza strains.

The research team looked at the genomes of influenza viruses to find a common target for possible vaccines. They also tested each vaccine for its efficacy against those viruses. Their models and clinical trials yielded mountains of data — much more than was typical in this type of study.

Matching patterns in study’s data

“There were multiple experiments, and there were multiple vaccines and multiple viruses,” Salge said. “With my background in MIS, Dr. Criado asked if I would take a look at it and help them organize the data, clean it up and kind of put it together for analysis, in addition to analyzing the data as well.”

Salge was able to automate the process of finding pattern matches in the data.

“I was looking at the different amino acid positions in the virus protein, and I didn’t know what they meant, for example, which regions were important versus those that were not and why,” Salge said. “But I found recurring patterns in specific positions. I noticed changes in notation in these positions too. So, I thought, ‘There’s something happening here,” and that’s what I told Dr. Criado.”

The patterns that Salge uncovered turned out to be part of the H5N1 genome that routinely mutated, making vaccines less effective. It helped to shed light on the path that future vaccine development should take.

“Dr. Criado and our co-author, Dr. David Swayne, said they wouldn’t have thought of those positions initially, but the patterns were so robust that we all started looking into it,” she said. “With what is known about sequencing and mutation and viruses in general, they realized that the patterns did make sense, from a theory perspective. And so our research highlighted the important genome regions to be examined in future work.”

The collaboration also showed Salge the power of data discovery and its value to laboratory research and other scientific fields.

“I think this is the beauty of interdisciplinary scientific work,” Salge said. “When you bring in researchers from different fields and with different perspectives and knowledge together, you can do amazing things. It was a very humbling experience.”