Deep learning-based phenotyping of crop seed banks and herbaria

点击翻译按钮获取中文摘要

Biodiversity underpins long term sustainability of agriculture as well as ecosystems. Current commercialcrops have a extremely narrow genetic base, which is a vulnerability. Seed banks and Herbaria representa concentrated resource, collected over many decades and include many heritage varieties with potentialvalue in the face of climate change and other pressures. The challenge is to extract this information in asystematic manner that useful to researchers, breeders and agriculture.We hypothesize that these physical archival materials can be mined for variation in useful traits usingnon- or minimally destructive approaches, that include image-based phenotyping, and this variationrelated to the underlying genetics.To test this, well documented grass and cereal collections are available in the AberInnovation SeedBiobank and in the IBERS' Crop Herbarium. The specific objectives are to:Obj1. Machine learning toolsTo develop robust deep learning protocols and test, a set of manually labelled images from selected testgrasses will be used to train neural networks to recognise and measure features. We have recentlyapplied deep learning to identify and count organ from both vouchers and uCT scanning (Masters project,Fig). These deep learning networks will be extended to quantify a wider variety of traits at differentlevels of image resolution. For example, higher resolution imaging and deep learning will be applied tomicroscopic features such as stomata and vessels that support water transport and gaseous exchangeprocesses and other minimally destructive modalities will be evaluated (X-Ray Fluorescence imaging forelemental analyses, metabolomics for nutritive content, etc).Obj2. De novo test-case experimentsTo formally address whether variation in these traits can be linked to underlying genetic variation (andtherefore be useful to breeders), the DR will use vouchers made from genetically unstructuredpopulations that have been previously created by crossing selected pairs of oats and of ryegrass. Thesemorphologically diverse and genetically characterised populations provide a rigorous test-case for validation of the approach. The student will build quantitative models to account for the genotypiccontribution to variation in features contributing to yield and quality, and compare to models producedusing conventional data.Obj3. Phenotyping the grass and oat herbariaA very extensive collection of grass species is populated by original wild samples (with geographiccoordinates) as well as duplicates grown locally under defined agronomy. These will be imaged and,where appropriate, uCT scanned for seed and stem traits. The samples are crossed referenced tobreeding ledgers and in many cases their pedigree can be traced through to current commercial varietieswhile seed can be recovered and regrown as necessary. This physical collection of >10000 samplesprovides an unexploited resource to examine genetic and environmental effects across time. The studentwill focus on the ryegrass and oats within the collection aiming to exploit advanced genetic analyticaltechniques such as GWAS.Justification and Likely Outcomes: This crop-focused project brings together the Phenomics Centre andSeed Bank at Aberystwyth with Computer Sciences in Surrey who have previously worked with KewGardens on biodiversity. This combination of expertise combined with large datasets from geneticallydefined species will allow design of optimised approaches to extract and interpret the informationcontent of more extensive collections, available in Botanical Gardens from across the world, andultimately extending beyond the grasses to other crops and into wider questions associated withbiodiversity.