Skip to Navigation Skip to Page Content

Research with corn phenotypes used to model data network

“The University of Missouri is one of the few places in the country that does interdisciplinary research well,” said Toni Kazic, the C.W. LaPierre professor in computer science. “It adds richness to the research, makes it easier to find solutions, and fosters a dialogue that will be ever richer’” added Kazic, who serves on the MU Informatics Institute (MUII) faculty and on the graduate faculties of the Interdisciplinary Plant Group, genetics, and molecular biology. She is delighted with the new commitment the University has made to informatics.

Kazic believes the actualization of MUII typifies the way science is being done today.  “So many problems in biology, medicine and agriculture revolve around complex systems with millions of interacting parts. Computation is essential to solving these, and the scale and complexity of the systems poses important new challenges for computer science,” she said.

Kazic’s research allows her to experience both ends of bioinformatics, planting, pollinating and collecting data from the same corn whose physical characteristics and genetic information are the foci of her work.

Before she got hooked on maize, she devoted herself to computational experimentation.

“I got very interested early on in how the networks of biochemical reactions in cells are organized and function,” Kazic said.

Understanding them required data on real biological systems, a need that led Kazic and her colleagues to work on how to represent biochemical information in databases so algorithms could reason with it, and on how databases could precisely share information.

In order to tackle the problem, she began genetic and computational experiments with the phenotypes—observable characteristics—of 50 mutations in corn that produce spots on leaves similar to those produced by a fungal infection, though no infection exists. At least half of the mutations, known as lesion mimics, are genetically distinct.

“When you look at the lesions from the different mutations, you find that there are many differences in the distribution of the spots, their size, and when they appear in the plant’s development,” said Kazic. “Two questions occur: What is the network of genes and reactions that produce these very complex phenotypes, and what algorithms could infer that network?”

Kazic and her student assistants use photography to document phenotypes of mutant corn leaves from plants that have been carefully back crossed to inbred lines of corn at the University’s South Farm. This is necessary to remove other mutants so that only the effects of lesion mutants can be measured.

In the field, waterproof barcodes are assigned to each row and to each plant. These denote the year, crop, genetic family, row, and plant number, uniquely identifying each plant. PDAs with built-in scanners are used to record the barcodes and collect the phenotypic and genetic data that will be linked to its genes in databases.

Identical protocols are used to pollinate each plant, and to photograph their leaves. These careful processes allow the researchers to make inferences about the functions of genes in the network. The portion of each plant’s genetic information that is involved in the lesions—a network of at least 200 genes—is matched to its phenotype and computationally modeled using parameters derived from the photographs. Inferences from the resulting data may inform research for a number of additional organisms.

“The next big challenge is for us to design algorithms that mimic the reasoning of geneticists studying small systems and see how well it scales to much larger systems,” said Kazic.

“As it turns out, progress on the mimic lesion phenotypes will help to understand the biology of southern and northern leaf blight in maize,” said Kazic. Southern leaf blight devastated the U.S. corn crop in the 1970s, and affects crops today. Many of the biological features of those diseases are analogous to the lesion mimic mutants, so we hope our computational models will be useful there, too.”

Kazic believes that the computational biochemical model may even be leveraged to help unravel systems within the brain, possibly revealing mechanisms of diseases such as Alzheimer’s.

“The maize community has been so supportive,” Kazic said of MU’s world-renowned maize research program. “We couldn’t do this without them.”