OSU integrative biology team uses machine learning in newly published Ebola virus research

Monday, September 25, 2023

Media Contact: Elizabeth Gosney | CAS Marketing and Communications Manager | 405-744-7497 | egosney@okstate.edu

A team of researchers led by Oklahoma State University integrative biology assistant professor Dr. Patrick Stephens recently unveiled critical insights into the transmission of the Ebola virus. 

The research team, which also included researchers from the University of Georgia and the Cary Institute of Ecosystems Studies, published the first comprehensive study of the potential host range of Ebola viruses including every mammal species in Africa. This study was led by Dr. Mekala Sundaram, who was a postdoctoral scholar in the Stephens Lab at OSU when it was published.

For years, the question of which animals were the natural hosts — or "reservoirs" — for the Ebola virus had baffled scientists, Stephens explained. It was previously known that Ebola viruses can infect various wild mammal species, including primates like chimpanzees, but primates tend to succumb quickly to infection, making them unlikely primary reservoirs. So, the research focused on identifying these natural reservoir species that ultimately transmit the virus to humans, either directly or through other animals.

"This is really important because species that are both likely to develop infections when exposed to the virus and tolerate infections well are the best candidates for primary reservoir species,” Stephens said. “These species will survive long enough to spread the virus to many other species. Our models appear to be able to predict which species will succumb quickly to or tolerate infections with about 90% accuracy."

The research addressed this enigma by conducting a comprehensive study encompassing over 500 species of wild mammals in Africa. No previous study included more than 120. Additionally, the study explored the role of the Newman Pick or NPC1 gene, which is known to affect Ebola virus infection in host cells, across multiple mammal species. Using machine learning, the team developed a model based on ecological and morphological traits to predict which species are likely to be infected.  

The research strongly implicated fruit bats in the family Pteropodidae as the primary reservoirs of Ebola viruses that maintain the virus in the wild, underlining their collective significance. An analysis of NPC1 sequences also proved especially promising; it was able to identify with 100% accuracy bat species in which infections have been previously documented in the wild. This suggests that NPC1 could be a critical genetic marker to consider in future studies of Ebola virus susceptibility in other species.

Using their findings from this research, Stephens’ team has several follow-up studies currently underway. One study, also led by Sundaram, is in review in an academic journal. It employs a statistical method called path analysis to explore how mammal characteristics interact with factors like fruit abundance in shaping transmission risk to human populations. Prior findings suggest that regions abundant in ficus fruits, such as wild figs, pose higher risks due to their attraction of fruit bats and primates, which are intermediaries in disease transmission to humans. 

“It’s kind of a perfect storm of factors," Stephens said. "Fruit bats and primates go there to eat fruit, and humans go there both to eat the fruit and to hunt bats and primates for meat. Bats are the ones spreading it to other species, and humans get it either directly from the bats or from primates infected by bats.” 

Members of the research team included Drs. John Paul Schmidt and John Drake from the University of Georgia, Dr. Barbara Han from the Cary Institute of Ecosystems Studies and Sundaram, who recently joined the faculty at UGA. 

"By soliciting and synthesizing the expertise of other team members, I conceived the idea for analyses, collected data, developed the models using appropriate machine learning methods, and led the writing of this paper for publication. I coordinated writing efforts among all team members," Sundaram said.

This OSU-UGA collaborative research was conducted as part of an NIH-funded project, "Spillover of Ebola and other Filoviruses at Ecological Boundaries," with total funding of $2.4 million. It will continue for nearly three more years.

“Ultimately there are two potential broad impacts of our work. First is a better understanding of how Ebola viruses are getting into human populations, and how they are being maintained in the wild.  However, ultimately we hope that the research will help prevent future outbreaks by identifying what regions and what human behaviors are likely to be high risk,” Stephens said.

Read the full research article, “Traits, phylogeny and host cell receptors predict Ebolavirus host status among African mammals” on the PLOS Biology website. You can find a link to this and Stephens’ and Sundaram’s other recent publications here.

Story By: Allie Putman, CAS Graduate Student | allie.putman@okstate.edu