Integrating machine learning to elucidate the genetic basis of 30 traits in G. barbadense BMC79 cultivar with superior fiber properties

With economic development and rising living standards, the demand for high-quality cotton fiber is increasing. Understanding the genetic basis of key traits for cotton with super fiber is crucial for breeding new cultivars. Here, we conducted QTL mapping and candidate gene identification for 30 important agronomic traits in the early-maturing, high-quality fiber cultivar Gossypium barbadense BMC79. We crossed BMC79 with upland cotton XLZ14 to generate an F2 population of 303 families and constructed a genetic map spanning 4026.30 cM with an average inter-bin distance of 0.31 cM. QTL analysis, integrated with machine learning, identified 55 QTLs for yield, fiber quality, and growth period related traits. Notably, QTLs including genes for fiber length (A01), lint percentage (A06), plant height (D11), fiber strength (D11), fiber uniformity (D12), and early maturity (D07), showed high phenotypic variance explained. Machine learning predicted several key candidate genes, such as Gh_A01G162500 (fiber length), Gh_A06G112000 (lint percentage), Gh_D11G351100 (fiber strength), and Gh_D07G112500 (flowering time). Importantly, Virus-Induced Gene Silencing (VIGS) validation showed that silencing Gh_D11G351100 significantly reduced fiber strength and length, confirming its role in fiber development. Our study provides valuable insights into the genetic basis of high-fiber-quality cotton varieties and offers important targets and references for the development of new cotton cultivars.