A Jupyter-based toolkit for teaching computational analysis and visualization to Food Science students

The goal of this project was to develop and deploy Jupyter-based learning tools for practical computational training in the Food Science curriculum. We developed interactive e-notebooks and a pre-configured JupyterHub-based computing environment for food informatics where students can perform various the analyses for bioinformatics and data analysis with regard to food sciences. This was motivated by a need to incorporate more active computational exercises into the food science curriculum. With support of LET, we set up a JupyterHub environment with custom installation of software needed (Goal 1). We further developed our e-notebooks for use in the «Applied Bioinformatics» course and deployed this on JupyterHub for streamlined integration with Moodle and use in the classroom (Goal 2). We also developed freestanding JupyterBooks for Python programming (Goal 3) and data visualization for Food Scientists (Goal 4), designed to be implemented as a self-paced online course, in a block course, or alongside others courses.

The materials developed in this course were first implemented in the Autumn Semester 2022 course, «Applied Bioinformatics: Microbiomes». Students used Jupyter notebooks deployed on JupyterHub to complete in-class exercises, homeworks, and group projects. Other materials developed in this project will be implemented in the same course in Autumn semester 2023, and as a self-paced course on python programming. The use of JupyterHub was a significant success in this course, eliminating hardware and installation issues for the students. Use of Jupyter notebooks also enabled reproducible data analysis, and notebooks that could be shared between students when working on the group project.

All project goals were achieved. One deviation was that we originally planned to create a JupyterBook from the Jupyter notebooks (Goal 2), but after initiating the project and working with JupyterHub we realized that JupyterHub is more effective for delivery of those course materials in a classroom setting and focused our efforts on improving the quality of those notebooks and their integration with JupyterHub and Moodle, instead of developing a JupyterBook. We evaluated the effect of JupyterHub on student learning by comparing student evaluations and feedback on the course between 2022 (with JupyterHub) and 2021 (preceding the use of JupyterHub), as well as our own experiences teaching the course in these years, and informal feedback from students. Overall, students and instructors had a positive impression of the course and materials. There were some significant lessons (with feedback from students) in the first round of using these materials: e.g., using GitHub to share notebooks creates some challenges for students unfamiliar with GitHub (e.g., due to merge conflicts). The positive lessons learned include the benefits of JupyterHub for «easy» deployment of computational notebooks for students to use in a classroom setting, and we already plan to expand use of JupyterHub for other classes in the Food Science curriculum. Lessons for improvement include figuring out better ways for deploy shared notebooks with version control systems, e.g., GitHub — our current plan is to improve documentation and training of students when using JupyterHub and GitHub together, to develop a list of FAQs and guidelines. This project already enabled enrollment of a higher number of students; increasing the capacity to 30 students in 2022. With further improvements, the capacity could in theory be increased further.

Jose Guanter Fonds