Authentic Research Experience and "Big Data" Analysis in the Classroom: Maize Response to Abiotic Stress

Abstract

Integration of inquiry-based approaches into curriculum is transforming the way science is taught and studied in undergraduate classrooms. Incorporating quantitative reasoning and mathematical skills into authentic biology undergraduate research projects has been shown to benefit students in developing various skills necessary for future scientists and to attract students to science, technology, engineering, and mathematics disciplines. While large-scale data analysis became an essential part of modern biological research, students have few opportunities to engage in analysis of large biological data sets. RNA-seq analysis, a tool that allows precise measurement of the level of gene expression for all genes in a genome, revolutionized molecular biology and provides ample opportunities for engaging students in authentic research. We developed, implemented, and assessed a series of authentic research laboratory exercises incorporating a large data RNA-seq analysis into an introductory undergraduate classroom. Our laboratory series is focused on analyzing gene expression changes in response to abiotic stress in maize seedlings; however, it could be easily adapted to the analysis of any other biological system with available RNA-seq data. Objective and subjective assessment of student learning demonstrated gains in understanding important biological concepts and in skills related to the process of science.

Figure 1.

The flow of an RNA-seq experiment. The steps shown in blue were performed by students. Students completed learning exercises only for the steps shown in green.

Figure 2.

Students’ prior exposure to the data analysis approaches used in the lab. During the first week of class, students were asked to rank their prior experience to gene expression analysis, analysis of large data sets, and using R or other programming tools in data analysis and data visualization. Data shown are for 85 students in a genetics course who completed this survey in 2014. The actual number of students is designated for each answer choice.

Figure 3.

Phenotypic effects of exposure to abiotic stress observed by students. B73 seedlings show strong response to cold stress with dry necrotic leaf edges and tips, while Mo17 seedlings show only minimal response to cold. Both B73 and Mo17 seedlings show response to heat stress with wilted leaves.

Figure 4.

Examples of graphs students used to visualize data and answer the questions. (A and B) Comparison of variation between two replicates of the same condition and between stress and control conditions. Log₂RPKM values are graphed for all maize genes. (C) The conservation of stress response. Many genes up-regulated in response to cold stress in B73 are also up-regulated in response to cold stress in Mo17, while many genes show response in only one of the genotypes. (D) The proportion of all maize genes, genes up-regulated in response to cold, and genes down-regulated in response to cold is shown. SE is shown with error bars. Three gene ontology categories significantly overrepresented among genes up-regulated in response to cold stress are shown (p < 0.05). (E) Abiotic stress exposure results in up- or down-regulation for a number of maize genes in each genotype. The Z-normalized RPKM values for all differentially expressed genes were used to perform hierarchical clustering of the gene expression values. The genotypes (B73: B; Mo17: M) and conditions (heat: red; control: green; cold: blue) are indicated below each column. Three replicates of each condition are shown. (F) Genes affected by cold stress are frequently up-regulated in response to heat stress as well. Genes up- and down-regulated for cold stress in B73 are shown, as is their response to heat stress. ND: the genes with no differential expression. The number of genes in each category is shown.

Figure 5.

Assessment of student learning. Student learning was assessed using a test consisting of 22 multiple-choice questions. Questions were separated into three categories, and the average proportion of correct answers for the questions in these categories was calculated for two courses (Principles of Genetics and Applied Biotechnology). Vertical bars show SD. For all three categories, the differences between pretest and posttest were significant as tested by paired t test (p < 0.01).