In this Computomics Podcast episode, we launch the new “Make Sense of Science” explainer series, a format designed to break down complex scientific and technical topics into clear, accessible conversations.
In this first episode, Computomics machine learning scientist Alaukik Saxena introduces deep learning for genomic prediction in plant breeding. He explains how breeders use DNA marker data, field data, and environmental information to predict important plant traits such as yield, flowering time, or plant height, often before new plants are even grown. The episode also looks at why this is such a challenging task: breeding data is complex, environments differ greatly, and models can easily give overly optimistic results if they are not tested carefully.
Alaukik also explains the difference between classical statistical models, machine learning, and deep learning, and where each approach has its strengths. A key message of the episode is that good models depend on good data. Better phenotyping, smarter experimental design, and realistic validation are just as important as the algorithms themselves.