A combination of ANNs and multivariate sensitivity analysis unveils critical factors to increase androgenesis efficiency in tomato (Solanum lycopersicum)

This study aimed to enhance androgenesis efficiency protocols by identifying critical factors using artificial neural networks and multivariate sensitivity analysis. Androgenesis studies were conducted on donor tomato plants grown under different lighting regimes (LEDs and natural light). Results indicated that LED lighting significantly enhanced photomorphogenesis, photosynthesis, but also subsequent anther development being crucial for successful haploid regeneration. An androgenesis model was developed using 60 different treatments, including as inputs: genotype, lighting regime, anther stage, and medium composition. Artificial Neural Networks (ANNs) and multivariate sensitivity analysis successfully ranked the importance of those four factors, identifying plant growth regulators (PGRs) media formulation as the most critical factors. The study also confirmed spontaneous haploid genome duplication in regenerants and validated the use of flow cytometry and cytogenetic analysis for ploidy determination. These findings provide valuable and practical insights into the factors driving androgenesis, facilitating the development of more efficient, targeted and cost-effective breeding strategies, such as the production of doubled haploid tomato.Key message: Our results, utilizing ANNs and multivariate analysis, identify key factors controlling androgenesis in tomato, leading to the development of innovative procedures for a rapid, cost-effective, and efficient production of doubled haploids. [ABSTRACT FROM AUTHOR]