Bioinformatics at the population level of biological organization: development of a methodology for estimating classical genetic parameters from multivariate categorical data

MACAT, PAULA; KOVALEVSKI, LEANDRO; CACCHIARELLI, PAOLO; QUAGLINO, MARTA; PRATTA, GUILLERMO R.

Mar del Plata

Congreso; IX Argentinian Congress of Bioinformatics and Computational Biology 9CAB2C; 2018

Asociación Argentina de Bioinformática y Biología Computacional - A2B2C

Bioinformatics has been extensively applied at the molecular level of biological organization but challenges are still a vacancy area for other more complex hierarchy of life, such as the population level. In fact, natural evolution and plant and animal breeding are emergent properties of populations. Classical Quantitative Genetics developed sophisticated mathematical and statistical models for analysing the inheritance of complex traits mostly based on parametric tests such as ANOVA, ANCOVA and correlation. Presently, molecular biology technics allow gaining high amounts of categorical data, for instance: presence or absence of polypeptides at a given tissue, which become true complex traits when submitted to integrative analysis. In previous reports, we informed advances in the development of a methodology for estimating the classical parameters average and variety heterosis, provided by the classical mating design of diallel crosses for phenotypic traits, to the polypeptide profiles from the pericarp at different ripening stages of tomato fruits harvested on homozygous and heterozygous genotypes. This approach was based on PerMANOVA test, a multivariate inference analysis suitable for dichotomic data. In the present communication, we report the estimation of variety effect and specific heterosis from the same database, which resulted statistically significant among the five parents and their 10 diallel hybrids. From a biological and agronomical viewpoint, this finding implies that polypeptide expression along tomato ripening is regulated by both additive and non-additive (dominance and epistasis) gene actions, so that if a given polypeptide pattern is desired, artificial selection followed by hybridization among pure lines should be achieved. From a Bioinformatics and Statistics viewpoint, we demonstrated that non parametrical model could be applied for estimating classical genetic parameters for categorical data generated by molecular technics.