The multidisciplinary team from Purdue University, University of Missouri, and University of Arkansas worked and delivered results on the project “Building Infrastructure and Connectivity for Small and Medium Scale Processing of Soy-Based Value Added Products: A Multistate Approach”. The team accomplished the stated deliverables of the project that would be of great help to the small and medium scale processors. Researchers at USDA-ARS at Missouri and West Lafayette produced a variety of seed for small and medium scale processing in the 2023 and 2024 growing seasons. This included 60 lbs of the high oleic experimental line and the commodity parent control, as well as smaller quantities of seed for approximately 50 novel high protein and low raffinose-stachyose containing mutants. The seed was characterized for protein and oil content and carbohydrate composition. This seed will provide a variety of potential composition to use in development of downstream methods for laboratory and small-scale processing, and the understanding we gain regarding the performance of these soybeans in food production pipelines guides our further germplasm development. The novel soybean line has been have reduced off-flavors and anti-nutrient. Research conducted at University of Missouri revealed that the Super line (HOLL with stachyose, raffinose, and lipoxygenase null) produced soymilk with low volatile concentrations compared to the commodity and HOLL commercial lines. Researchers at Purdue have characterized sugar profile for soybean varieties to assess their suitability for soymilk and tofu production. The small and micro-scale soy milk and tofu testing has been completed for soybean varieties. Overall, we were able to successfully conduct a comprehensive compositional phenotype for a diverse set of soybean samples. Additionally, great success was made in establishment of micro and small-scale soymilk and tofu production tests that relate well to soybean quality.
The team have developed computational multi-scale models to study and correlate soybean seed composition to soy-based product mechanical properties (like tofu made from soy gel) which would be beneficial to quickly determine the optimal seed type to plant. We find three distinct clusters of seed types from the data provided by Missouri: seeds that produce softer gels, seeds that produce firmer gels, and strong seeds that are more resistant to cracking. We evaluated the results with machine learning models and developed a model that can classify the seed types with 99% accuracy and find that linolenic, linoleic, and palmitic fatty acids have the most effect on predicted soy gel porousness. We then developed a multiscale analysis to quantify the effect of extrusion on soy gel texture and firmness, which would be beneficial in identifying the optimal processing parameters for tuning the extruded texture. Results showed that softer gels extrude more linearly and with less defects.
The team have also developed a reproducible lab-scale protocol to prepare soybean meal pellets (low fat and high fat pellets) and oil using mechanical screw oil/press requiring only 500 g sample sizes. Soybean meal pellets and oil from four unique varieties were prepared using this protocol and the quality of the generated coproducts were investigated. The developed protocol was highly repeatable, and generated coproducts retained all the nutritional characteristics associated with the raw soybean kernels. In developing alternative method to prepare soy protein ingredient, researchers have utilized high intensity ultrasound to isolate protein from ground soybean. The optimum processing condition leading to maximum yield and lowest energy was identified. In another study, researchers compared different extraction methods to isolate protein from mechanical-press soybean cake. This study demonstrated how different methods influence yields and properties of isolated protein ingredients. The use of ethanol or isop