Modeling multiple phenotypes in wheat using data-driven genomic exploratory factor analysis and Bayesian network learning

Multi-omic data integration for the study of production, carcass, and meat quality traits in Nellore cattle

Data integration using hierarchical analysis based on the central dogma or common pathway enrichment analysis may not reveal non-obvious relationships among omic data. Here, we applied factor analysis (FA) and Bayesian network (BN) modeling to integrate different omic data and complex traits by latent variables (production, carcass, and meat quality traits). A total of 14 latent variables were identified: five for phenotype, three for miRNA, four for protein, and two for mRNA data. Pearson correlation coefficients showed negative correlations between latent variables miRNA 1 (mirna1) and miRNA 2 (mirna2) (−0.47), ribeye area (REA) and protein 4 (prot4) (−0.33), REA and protein 2 (prot2) (−0.3), carcass and prot4 (−0.31), carcass and prot2 (−0.28), and backfat thickness (BFT) and miRNA 3 (mirna3) (−0.25). Positive correlations were observed among the four protein factors (0.45–0.83): between meat quality and fat content (0.71), fat content and carcass (0.74), fat content and REA (0.76), and REA and carcass (0.99). BN presented arcs from the carcass, meat quality, prot2, and prot4 latent variables to REA; from meat quality, REA, mirna2, and gene expression mRNA1 to fat content; from protein 1 (prot1) and mirna2 to protein 5 (prot5); and from prot5 and carcass to prot2. The relations of protein latent variables suggest new hypotheses about the impact of these proteins on REA. The network also showed relationships among miRNAs and nebulin proteins. REA seems to be the central node in the network, influencing carcass, prot2, prot4, mRNA1, and meat quality, suggesting that REA is a good indicator of meat quality. The connection among miRNA latent variables, BFT, and fat content relates to the influence of miRNAs on lipid metabolism. The relationship between mirna1 and prot5 composed of isoforms of nebulin needs further investigation. The FA identified latent variables, decreasing the dimensionality and complexity of the data. The BN was capable of generating interrelationships among latent variables from different types of data, allowing the integration of omics and complex traits and identifying conditional independencies. Our framework based on FA and BN is capable of generating new hypotheses for molecular research, by integrating different types of data and exploring non-obvious relationships.

Dissecting Breeders' Sense via Explainable Machine Learning Approach: Application to Fruit Peelability and Hardness in Citrus

"Genomics-assisted breeding", which utilizes genomics-based methods, e.g., genome-wide association study (GWAS) and genomic selection (GS), has been attracting attention, especially in the field of fruit breeding. Low-cost genotyping technologies that support genome-assisted breeding have already been established. However, efficient collection of large amounts of high-quality phenotypic data is essential for the success of such breeding. Most of the fruit quality traits have been sensorily and visually evaluated by professional breeders. However, the fruit morphological features that serve as the basis for such sensory and visual judgments are unclear. This makes it difficult to collect efficient phenotypic data on fruit quality traits using image analysis. In this study, we developed a method to automatically measure the morphological features of citrus fruits by the image analysis of cross-sectional images of citrus fruits. We applied explainable machine learning methods and Bayesian networks to determine the relationship between fruit morphological features and two sensorily evaluated fruit quality traits: easiness of peeling (Peeling) and fruit hardness (FruH). In each of all the methods applied in this study, the degradation area of the central core of the fruit was significantly and directly associated with both Peeling and FruH, while the seed area was significantly and directly related to FruH alone. The degradation area of albedo and the area of flavedo were also significantly and directly related to Peeling and FruH, respectively, except in one or two methods. These results suggest that an approach that combines explainable machine learning methods, Bayesian networks, and image analysis can be effective in dissecting the experienced sense of a breeder. In breeding programs, collecting fruit images and efficiently measuring and documenting fruit morphological features that are related to fruit quality traits may increase the size of data for the analysis and improvement of the accuracy of GWAS and GS on the quality traits of the citrus fruits.