A maize plant. Photo by: Neil Palmer / CIAT
If you were given datasets with genetic information and yields from trials of different experimental maize hybrids, along with information on soil conditions and recorded weather of a U.S. growing region, how will you predict the performance of these hybrids?
And what if the genetic information dataset included an average of 12,000 genetic markers or unique DNA sequences to identify each hybrid?
Those were the questions that the CIAT data analytics team needed to address as part of an international competition organized by agribusiness giant Syngenta in partnership with the Institute for Operations Research and the Management Sciences (INFORMS), the international society of operations research, management science, and analytics practitioners.
The Syngenta Crop Challenge in Analytics began in 2016. Since then, dozens of teams of data experts from around the globe have participated in the competition.
The CIAT data analytics team — including Andres Aguilar, Sylvain Delerce, Hugo Andres Dorado, Michael Caraccio, Juan Camilo Rivera, Maria Camila Gomez, Steven Humberto Sotelo, and Anestis Gkanogiannis — proposed a solution that can speed up maize hybrid breeding using machine learning.
The proposal proved to be among the five best entries to the 2018 Syngenta Crop Challenge in Analytics.
The selection followed a peer review process by an INFORMS panel of experts of diverse technical background. The process considered a set of criteria that includes the proposal’s rigor, clarity, and innovation.
“This is a dream come true,” said team member Juan Camilo Rivera.
According to Rivera, the solution, though proposed for the competition, can be applied anywhere and to any crop. One only needs to have the genetic information of that crop, as well as the weather data and soil conditions of a particular area.
Breeders specifically, he noted, can use the proposed tool.
“Like others, this tool can still be improved,” added Rivera, a mathematician. “But I believe it can lead to good results, especially for crop breeding.”
Representatives from the team will need to make a 20-minute presentation of the proposal at the 2018 INFORMS Business Analytics Conference later this month in Baltimore, Maryland, USA.
A 10-minute Q&A by the judges will follow.
From the five finalists, three will get prizes.
Dr. Daniel Jimenez, who supervises the team, said he “pushed [his team] hard to do this.” This, he added, highlighted that CIAT “is not just doing good science but also most importantly good science for development.”
The work that Jimenez and his team are doing has already received international recognition. In 2017, the United Nations Framework Convention on Climate Change selected the climate adaptation project that they and other colleagues as CIAT and the CGIAR Research Program on Climate Change, Agriculture, and Food Security have in Latin America as a winner of the 2017 Momentum for Change Lighthouse Activities.
“I am proud of what we’ve achieved so far,” Jimenez said. “It’s going to be tough, given the caliber of the other four finalists. And that’s also why just making it to the final stage of the 2018 INFORMS Business Analytics Conference is already a huge recognition.”