To explore a robust tool for advancing digital breeding practices through an artificial intelligence-driven phenotype prediction expert system, we undertook a thorough analysis of 11 non-linear regression models. Our investigation specifically emphasized the significance of Support Vector Regression (SVR) and SHapley Additive exPlanations (SHAP) in predicting soybean branching. By using branching data (phenotype) of 1918 soybean accessions and 42 k SNP (Single Nucleotide Polymorphism) polymorphic data (genotype), this study systematically compared 11 non-linear regression AI models, including four deep learning models (DBN (deep belief network) regression, ANN (artificial neural network) regression, Autoencoders regression, and MLP (multilayer perceptron) regression) and seven machine learning models (e.g., SVR (support vector regression), XGBoost (eXtreme Gradient Boosting) regression, Random Forest regression, LightGBM regression, GPs (Gaussian processes) regression, Decision Tree regression, and Polynomial regression). After being evaluated by four valuation metrics: R² (R-squared), MAE (Mean Absolute Error), MSE (Mean Squared Error), and MAPE (Mean Absolute Percentage Error), it was found that the SVR, Polynomial Regression, DBN, and Autoencoder outperformed other models and could obtain a better prediction accuracy when they were used for phenotype prediction. In the assessment of deep learning approaches, we exemplified the SVR model, conducting analyses on feature importance and gene ontology (GO) enrichment to provide comprehensive support. After comprehensively comparing four feature importance algorithms, no notable distinction was observed in the feature importance ranking scores across the four algorithms, namely Variable Ranking, Permutation, SHAP, and Correlation Matrix, but the SHAP value could provide rich information on genes with negative contributions, and SHAP importance was chosen for feature selection. The results of this study offer valuable insights into AI-mediated plant breeding, addressing challenges faced by traditional breeding programs. The method developed has broad applicability in phenotype prediction, minor QTL (quantitative trait loci) mining, and plant smart-breeding systems, contributing significantly to the advancement of AI-based breeding practices and transitioning from experience-based to data-based breeding.

Food security is always a pressing agenda worldwide. The grain production in many areas has decreased due to the reduction in agricultural research funding and infrastructure investment. In this paper, we employed the Extreme-Tree algorithm to determine the main effectors in grain production in Hexi Corridor, Gansu, China, during 2002–2018. First, we applied the three-stage super-SBM DEA to precisely assess agricultural production. Then, we used the Extremely randomized trees algorithm to quantify the importance of each factor. Our results show that the variant of average efficiency score at the first stage was minimal. After removing the influence of environmental factors on production efficiency, the more accurate efficiency score was decreasing from 2002 to 2018. The R² value of the Extra-Tree model was 0.989 in the grain production analysis. Our research shows that grain production in the Hexi Corridor was controlled by human-driven but not nature-driven during our research period. Based on the importance attribution analysis of each model, it showed that the importance of human-driven investment occupied 93.7% of grain production. The importance of nature-driving was about 6.3%. Accordingly, we proposed corresponding opinions and suggestions to government and growers.

Cassava plays an important role in the life and diet of several communities worldwide. It is used in various processed forms and has become the focus of research and innovation in recent years. Bioprocessing techniques, such as fermentation, have been utilized to create new avenues for its application in food and functional products. The fermentation of cassava can enhance its nutritional value, improve its shelf life, and increase its digestibility. However, the effect of salt-mediated fermentation on microbial diversity and potential benefits has not been widely reported. In this study, the effect of six levels of salt (NaCl), ranging from 0 to 25% after 50 days of spontaneous cassava fermentation at 25–30 °C, was investigated. A total of 30 bacterial isolates were selected for molecular characterization. A proprietary pool of universal 16S rRNA primers and species-specific primers were used to amplify a wide variety of species in clonal samples. The bacteria observed include Lactiplantibacillus, Limosilactobacillus, and Weissella. The population of Lactiplantibacillus in the cassava microflora increased with and without salt treatment, while Weissella, a new genus, was detected in 20 and 25% of salt-treated samples. Lactiplantibacillus and Weissella accounted for 32 and 68% and 60 and 40% of the bacterial populations detected at 20 and 25% salt treatment, respectively. A total of 17 strains and sequences were identified from the 30 isolates screened. Sequencing results classified the 30 isolates into four groups, of which 76.67% were Limosilactobacillus. Phylogenetic analysis showed that all 17 strains were divided into three clusters. These results show that high salt-mediated fermentation of cassava can trigger a shift in dominance in the bacterial community, changing the diversity of the microbial community.

Fruit allergens are proteins from fruits or pollen that cause allergy in humans, an increasing food safety concern worldwide. With the globalization of food trade and changing lifestyles and dietary habits, characterization and identification of these allergens are urgently needed to inform public awareness, diagnosis and treatment of allergies, drug design, as well as food standards and regulations. This study conducted a phylogenetic reconstruction and protein clustering among 60 fruit and pollen allergens from 19 species, and analyzed the clusters, in silico, for cross-reactivity (IgE), 3D protein structure prediction, transmembrane and signal peptides, and conserved domains and motifs. Herein, we wanted to predict the likelihood of their interaction with antibodies, as well as cross-reactivity between the many allergens derived from the same protein families, as the potential for cross-reactivity complicates the management of fruit allergies. Phylogenetic analysis classified the allergens into four clusters. The first cluster (n = 9) comprising pollen allergens showed a high risk of cross-reactivity between eight allergens, with Bet v1 conserved domain, but lacked a transmembrane helix and signal peptide. The second (n = 10) cluster similarly suggested a high risk of cross-reactivity among allergens, with Prolifin conserved domain. However, the group lacked a transmembrane helix and signal peptide. The third (n = 13) and fourth (n = 29) clusters comprised allergens with significant sequence diversity, predicted low risk of cross-reactivity, and showed both a transmembrane helix and signal peptide. These results are critical for treatment and drug design that mostly use transmembrane proteins as targets. The prediction of high risk of cross-reactivity indicates that it may be possible to design a generic drug that will be effective against the wide range of allergens. Therefore, in the past, we may have avoided the array of fruit species if one was allergic to any one member of the cluster.

Plant breeding is gaining importance as a sustainable tool to address the challenges posed by a growing global population and enhance food security. Advanced high-throughput omics technologies are utilized to accelerate crop improvement and develop resilient varieties with higher yield performance. These technologies generate vast genetic data, which can be exploited to manipulate key plant characteristics for crop improvement. The integration of big data and AI in plant breeding has the potential to revolutionize the field and increase food security. By using branching data (phenotype) of 1918 soybean accessions and 42k SNP polymorphic data (genotype), this study systematically compared 11 non-linear regression AI models, including four deep learning models (DBN regression, ANN regression, Autoencoders regression, and MLP regression) and seven machine learning models (e.g., SVR, XGBoost regression, Random Forest regression, LightGBM regression, GPS regression, Decision Tree regression, and Polynomial regression). After being evaluated by four valuation metrics: R2 (R-squared), MAE (Mean Absolute Error), MSE (Mean Squared Error), and MAPE (Mean Absolute Percentage Error), it was found that the SVR, ANN, and Autoencoder outperformed other models and could obtain a better prediction accuracy if they were used for phenotype prediction. To support the evaluation of deep learning methods, feature importance and GO enrichment analyses were conducted. After comprehensively comparing four feature importance algorithms, there was no significant difference among the feature importance ranking score among these four algorithms, but the SHAP value could provide rich information on genes with negative contributions, and SHAP importance was chosen for feature selection. The genes identified by the SVR model plus SHAP importance combination clearly grouped into three clusters on the soybean whole genome. Our GO enrichment results also confirmed the prediction accuracy of this methods combination. The results of this study offer valuable insights for AI-mediated plant breeding, addressing challenges faced by traditional breeding programs. The method developed has broad applicability in phenotype prediction, minor QTL mining, and plant smart-breeding systems, contributing significantly to the advancement of AI-based breeding practices and transitioning from experience-based to data-based breeding.