Mapping pea seed composition through strategic selection of accessions from the Nordic gene bank

This study aims to utilise natural variation in pea seed composition from NordGen collections to identify key traits for optimized plant-based ingredients functionality while minimizing refined extraction processes. Given the impracticality of chemically analysing 1942 accessions, an algorithm-assisted approach was employed, using image-derived features and datasets to pre-select 51 accessions. Protein content, thousand kernel weight, perimeter, and G-value were determined as primary criteria via PCA, capturing variations in protein composition and other key components. Protein and starch content ranged from 21.2 to 36.9 % and 21.0-48.1 %, respectively. Image analysis linked geometry to composition, aiding pea selection and application. X-ray scattering differentiates peas based on starch structure. Proteomic profiling revealed that legumin and vicilin varied most, with legumin dominant in smooth peas and vicilin in wrinkled ones, enabling control of their ratio through selection. This study highlights the potential of using natural variation of seed composition for less-refined plant-based ingredients for various applications.