Full text
| |
| Author(s): |
Alexandre Hild Aono
Total Authors: 1
|
| Document type: | Doctoral Thesis |
| Press: | Campinas, SP. |
| Institution: | Universidade Estadual de Campinas (UNICAMP). Instituto de Biologia |
| Defense date: | 2024-05-09 |
| Examining board members: |
Anete Pereira de Souza;
Maria Imaculada Zucchi;
Renato Vicentini;
Gabriel de Siqueira Gesteira;
Evandro Novaes
|
| Advisor: | Ana Carolina Lorena; Anete Pereira de Souza; Marcos Gonçalves Quiles |
| Abstract | |
Sugarcane plays a significant role in the global economy, both in sugar production and in generating bioenergy through its bagasse. However, understanding the genetic architecture related to its economically important traits has been limited due to its high genomic complexity and the lack of adequate statistical models to efficiently associate genotype and phenotype. In the developed thesis, innovative genetic models were created to explore the molecular mechanisms regulating economically important traits in sugarcane. Initially, a comprehensive characterization of the protein kinase family was conducted, investigating its chromosomal and allelic distribution, intronic organization, composition of protein domains, protein properties, among other aspects, in order to understand the genomic complexity of the species and its functional implications. Subsequently, genomic prediction models based on machine learning were developed for different sugarcane traits, ranging from discrete characteristics such as resistance to brown rust, to complex quantitative traits such as stem diameter and height. Using feature selection techniques and classification algorithms, relevant genomic associations for brown rust resistance were identified, establishing good practices for SNP identification and association evaluation. Additionally, a methodology called joint learning was developed, combining classification and regression algorithms with feature selection techniques, allowing not only the prediction of the genetic value associated with phenotypic characteristics but also the estimation of phenotypic performance groups, facilitating genotype selection in breeding programs. Finally, the genetic architecture related to sugar accumulation in sugarcane was investigated, using both conventional genomic association techniques and the developed machine learning models. This investigation included the analysis of associated gene expression, its relationship with sugar accumulation variations, and the integration of these findings through gene co-expression networks. The results of this study not only contribute to the theoretical understanding of genotype-phenotype association in sugarcane but also provide valuable tools for genetic improvement programs, opening new perspectives for investigating complex traits and identifying potential biotechnological targets for molecular improvement of this crop (AU) | |
| FAPESP's process: | 19/03232-6 - Genome wide selection in sugarcane using machine learning and complex networks for economically important traits |
| Grantee: | Alexandre Hild Aono |
| Support Opportunities: | Scholarships in Brazil - Doctorate (Direct) |