Interactive practicals on the data side of plant resilience. Each one runs entirely in your browser, with no installation and no coding required. Work through a practical first, then use its answer model to check your reasoning.
Lecture 1 · Breeding with markers
Use a GWAS to find the DNA marker behind a trait, then plan crosses and backcrosses to move it into an elite line. Read Manhattan plots and steer a breeding programme.
When many genes each nudge a trait, single markers are not enough. Train a model over all markers at once to predict a plant's breeding value, and see where it succeeds and fails.
Lecture 2 · Finding structure in expression data
Explore real RNA-seq data from infected Arabidopsis. Read a clustered heatmap, tune hierarchical clustering, and build a gene correlation network to find structure in expression data.
Simulate a gene regulatory network as equations. Pulse a drought, lengthen it, and knock out genes to see how the drought response trades off against a plant's defence.
Lecture 3 · Learning from images
Train a real neural network, in your browser, on photographs of tomato leaves. Meet a 90%-accurate model that is worthless, then watch a good one learn entirely the wrong thing.
Lecture 4 · From gene lists to crops
A cluster of genes is just a list of names until you know what they do. Look genes up in the Gene Ontology, then let a statistical test name the biological process a whole cluster shares.
A gene you found in Arabidopsis is no use to a farmer. Find the matching gene in wheat, rice or potato, work out why some genes have fifty matches and others none, and align the two proteins yourself.
These pages are the interactive, browser-based version of the practicals for the Introduction to Climate-Resilient Crops minor. They mirror the original Jupyter notebooks by Ben Noordijk but need no setup: everything, including the simulations, runs in JavaScript on your own machine. Built with the help of Claude Code.